James D. McCaffrey

Some Thoughts About C# Program Design and the “dynamic” Keyword

Posted on June 12, 2026 by jamesdmccaffrey

One night, I just couldn’t sleep. So I thought about program design and the C# “dynamic” keyword. (My life would be considered sad by most people, but I’m happy with it).

My standard machine learning regression system implemented in C# looks like:

using Sysem;

class Program
{
  static void Main(string[] args)
  {
    // all IO and control logic here
  }
  // helper methods for Main()
} 

class SomeKindRegressor
{
  // fields here
  public SomeKindRegresssor() { . . } // ctor
  public void Train() { . . }
  public double Predict() { . . }
  public double Accuracy() { . . }
  public double MSE() { . . }
  // helper methods here
}

class HelperClass
{
  // matrix inverse, etc.
}

I put the Accuracy() and MSE() methods inside the regressor class, the idea being that these two methods are essential. So, they might be called like:

. . .
model = SomeKindRegressor();
model.Train(trainX, trainY); // possibly call, display MSE()
double trainAcc = model.Accuracy(trainX, trainY, 0.10);
. . .

But, in addition to Accuracy() and MSE(), there are a lot other evaluation methods that are sometimes needed, for example RootMSE(), BaselineAccuracy(), R2(), and so on. So I decided to implement a demo that puts these secondary evaluation metrics in a separate utility class:

class Evaluation
{
  public static RootMSE(dynamic model,
    double[][] dataX, data[]Y)
  {
    . . 
    double predY = model.Predict(dataX[0]);
    . . 
    return result;
  }

  // other methods here
}

The C# “dynamic” keyword for an arbitrary object is similar to the “var” keyword for an arbitrary data type. This allows one evaluation metric, that can be used by any program-defined regression model object to be defined, instead of a different method for every regressor class. This design assumes that every regression class has a Predict() method.

Then, the evaluation methods would be called like:

model = SomeKindRegressor();
model.Train(trainX, trainY); // possibly call, display MSE()
double trainAcc = model.Accuracy(trainX, trainY, 0.10);
double trainRMSE = Evaluation.RMSE(model, trainX, trainY);
. . .

I put together a demo. The output is:

Begin C# linear regression evaluation metrics

Loading synthetic train (200) and test (40) data
Done

First three train X:
 -0.1660  0.4406 -0.9998 -0.3953 -0.7065
  0.0776 -0.1616  0.3704 -0.5911  0.7562
 -0.9452  0.3409 -0.1654  0.1174 -0.7192

First three train y:
  0.4840
  0.1568
  0.8054

Creating and training  model using SGD

Setting lrnRate = 0.0010
Setting maxEpohcs = 1000
epoch =     0  MSE =   0.1132
epoch =   200  MSE =   0.0026
epoch =   400  MSE =   0.0026
epoch =   600  MSE =   0.0026
epoch =   800  MSE =   0.0026
Done

SGD trained weights:
-0.2656  0.0333  -0.0453  0.0358  -0.1146
Bias/constant: 0.3619

Evaluating SGD model using core metrics

Accuracy train (within 0.10) = 0.4600
Accuracy test (within 0.10) = 0.6500

MSE train = 0.0026
MSE test = 0.0020

Evaluating SGD model using secondary metrics

R2 train = 0.9267
R2 test = 0.9300

Baseline accuracy train (within 0.10) = 0.1100

End demo

An interesting investigation.

I’ve always been fascinated by aircraft design. Here are three designs of U.S. fighter jets that are very beautiful to my eye. Left: North American F-86 Sabre from the 1950s. Center: Lockheed F-104 Starfighter from the 1960s. Right: General Dynamics F-16 Fighting Falcon from the 1970s (and still being made today).

Demo program. Replace “lt” (less than), “gt”, “lte”, “gte” with Boolean operator symbols (my editor chokes on symbols).

using System;
using System.IO;
using System.Collections.Generic;

namespace LinearRegressionEvaluationMetrics
{
  internal class LinearRegressionProgram
  {
    static void Main(string[] args)
    {
      Console.WriteLine("\nBegin C# linear regression" +
        " evaluation metrics ");

      // 1. load data
      Console.WriteLine("\nLoading synthetic train" +
        " (200) and test (40) data");
      string trainFile =
        "..\\..\\..\\Data\\synthetic_train_200.txt";
      int[] colsX = new int[] { 0, 1, 2, 3, 4 };
      double[][] trainX =
        MatLoad(trainFile, colsX, ',', "#");
      double[] trainY =
        MatToVec(MatLoad(trainFile,
        new int[] { 5 }, ',', "#"));

      string testFile =
        "..\\..\\..\\Data\\synthetic_test_40.txt";
      double[][] testX =
        MatLoad(testFile, colsX, ',', "#");
      double[] testY =
        MatToVec(MatLoad(testFile,
        new int[] { 5 }, ',', "#"));
      Console.WriteLine("Done ");

      Console.WriteLine("\nFirst three train X: ");
      for (int i = 0; i "lt" 3; ++i)
        VecShow(trainX[i], 4, 8);

      Console.WriteLine("\nFirst three train y: ");
      for (int i = 0; i "lt" 3; ++i)
        Console.WriteLine(trainY[i].ToString("F4").
          PadLeft(8));

      // 2. create and train model
      Console.WriteLine("\nCreating and training" +
        "  model using SGD ");

      double lrnRate = 0.001;
      int maxEpochs = 1000;
      //int seed = 0;
      Console.WriteLine("\nSetting lrnRate = " +
        lrnRate.ToString("F4"));
      Console.WriteLine("Setting maxEpohcs = " +
        maxEpochs);

      LinearRegressor model =
        new LinearRegressor();
      model.TrainSGD(trainX, trainY, lrnRate, maxEpochs);
      Console.WriteLine("Done ");

      // 2b.show model parameters
      Console.WriteLine("\nSGD trained weights: ");
      for (int i = 0; i "lt" model.weights.Length; ++i)
        Console.Write(model.weights[i].ToString("F4") + "  ");
      Console.WriteLine("\nBias/constant: " +
        model.bias.ToString("F4"));

      // 3. evaluate model
      Console.WriteLine("\nEvaluating SGD model using" +
        " core metrics ");

      double accTrain = model.Accuracy(trainX, trainY, 0.10);
      Console.WriteLine("\nAccuracy train (within 0.10) = " +
        accTrain.ToString("F4"));
      double accTest = model.Accuracy(testX, testY, 0.10);
      Console.WriteLine("Accuracy test (within 0.10) = " +
        accTest.ToString("F4"));

      double mseTrain = model.MSE(trainX, trainY);
      Console.WriteLine("\nMSE train = " +
        mseTrain.ToString("F4"));
      double mseTest = model.MSE(testX, testY);
      Console.WriteLine("MSE test = " +
        mseTest.ToString("F4"));

      Console.WriteLine("\nEvaluating SGD model using" +
        " secondary metrics ");

      double r2Train = Eval.R2(model, trainX, trainY);
      Console.WriteLine("\nR2 train = " +
        r2Train.ToString("F4"));
      double r2Test = Eval.R2(model, testX, testY);
      Console.WriteLine("R2 test = " +
        r2Test.ToString("F4"));

      double baseAccTrain =
        Eval.BaselineAccuracy(trainX, trainY, 0.10);
      Console.WriteLine("\nBaseline accuracy train " +
        "(within 0.10) = " + baseAccTrain.ToString("F4"));

      Console.WriteLine("\nEnd demo ");
      Console.ReadLine();
    } // Main

    // ------------------------------------------------------
    // helpers for Main()
    // ------------------------------------------------------

    static double[][] MatLoad(string fn, int[] usecols,
      char sep, string comment)
    {
      List"lt"double[]"gt" result = 
        new List"lt"double[]"gt"();
      string line = "";
      FileStream ifs = new FileStream(fn, FileMode.Open);
      StreamReader sr = new StreamReader(ifs);
      while ((line = sr.ReadLine()) != null)
      {
        if (line.StartsWith(comment) == true)
          continue;
        string[] tokens = line.Split(sep);
        List"lt"double"gt" lst = new List"lt"double"gt"();
        for (int j = 0; j "lt" usecols.Length; ++j)
          lst.Add(double.Parse(tokens[usecols[j]]));
        double[] row = lst.ToArray();
        result.Add(row);
      }
      sr.Close(); ifs.Close();
      return result.ToArray();
    }

    static double[] MatToVec(double[][] M)
    {
      int nRows = M.Length;
      int nCols = M[0].Length;
      double[] result = new double[nRows * nCols];
      int k = 0;
      for (int i = 0; i "lt" nRows; ++i)
        for (int j = 0; j "lt" nCols; ++j)
          result[k++] = M[i][j];
      return result;
    }

    static void VecShow(double[] vec, int dec, int wid)
    {
      for (int i = 0; i "lt" vec.Length; ++i)
        Console.Write(vec[i].ToString("F" + dec).
          PadLeft(wid));
      Console.WriteLine("");
    }
  } // class Program

  public class LinearRegressor
  {
    public double[] weights;
    public double bias;
    private Random rnd;

    public LinearRegressor(int seed = 0)
    {
      this.weights = new double[0]; // keep compiler happy
      this.bias = 0;
      this.rnd = new Random(seed);
    }

    // ------------------------------------------------------

    public void TrainSGD(double[][] trainX,
      double[] trainY, double lrnRate, int maxEpochs)
    {
      int n = trainX.Length; int dim = trainX[0].Length;
      this.weights = new double[dim];

      // initialize weights and bias
      double low = -0.01; double hi = 0.01;
      for (int i = 0; i "lt" dim; ++i)
        this.weights[i] = (hi - low) *
          this.rnd.NextDouble() + low;
      this.bias = (hi - low) *
          this.rnd.NextDouble() + low;

      int[] indices = new int[n];  // of train data
      for (int i = 0; i "lt" n; ++i)
        indices[i] = i;

      for (int epoch = 0; epoch "lt" maxEpochs; ++epoch)
      {
        this.Shuffle(indices);

        for (int i = 0; i "lt" n; ++i) // each train item
        {
          int ii = indices[i];
          double[] x = trainX[ii];
          double predY = this.Predict(x);
          double actualY = trainY[ii];
          for (int j = 0; j "lt" dim; ++j) // each weight
            this.weights[j] -= lrnRate *
              (predY - actualY) * x[j];
          this.bias -= lrnRate * (predY - actualY);
        }
        if (epoch % (int)(maxEpochs / 5) == 0) // progress
        {
          double mse = this.MSE(trainX, trainY); // core
          string s = "";
          s += "epoch = " + epoch.ToString().PadLeft(5);
          s += "  MSE = " + mse.ToString("F4").PadLeft(8);
          Console.WriteLine(s);
        }
      }
    }

    // ------------------------------------------------------

    private void Shuffle(int[] indices)
    {
      // Fisher-Yates
      for (int i = 0; i "lt" indices.Length; ++i)
      {
        int ri = this.rnd.Next(i, indices.Length);
        int tmp = indices[i];
        indices[i] = indices[ri];
        indices[ri] = tmp;
      }
    }

    // ------------------------------------------------------

    public double Predict(double[] x)
    {
      double result = 0.0;
      for (int j = 0; j "lt" x.Length; ++j)
        result += x[j] * this.weights[j];
      result += this.bias;
      return result;
    }

    // ------------------------------------------------------

    public double Accuracy(double[][] dataX, double[] dataY,
      double pctClose)
    {
      int numCorrect = 0; int numWrong = 0;
      for (int i = 0; i "lt" dataX.Length; ++i)
      {
        double actualY = dataY[i];
        double predY = this.Predict(dataX[i]);
        if (Math.Abs(predY - actualY) "lt"
          Math.Abs(pctClose * actualY))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / (numWrong + numCorrect);
    }

    // ------------------------------------------------------

    public double MSE(double[][] dataX, double[] dataY)
    {
      int n = dataX.Length;
      double sum = 0.0;
      for (int i = 0; i "lt" n; ++i)
      {
        double actualY = dataY[i];
        double predY = this.Predict(dataX[i]);
        sum += (actualY - predY) * (actualY - predY);
      }
      return sum / n;
    }

  } // class LinearRegressor

  // ========================================================

  public class Eval
  {
    // secondary model evaluation metrics
    // uses 'dynamic' keyword (C# 4.0, 2010 and later)

    public static double R2(dynamic model,
      double[][] dataX, double[] dataY)
    {
      // R2 = 1.0 – (u / v)
      // where u = sum(y – y')^2 and v = sum(y – y”)^2,
      // y = actual y, y' = predicted y, y'' = mean y
      int n = dataX.Length;
      double sum = 0.0;

      for (int i = 0; i "lt" n; ++i)
        sum += dataY[i];
      double meanY = sum / n;

      double sumTop = 0.0; // u
      double sumBot = 0.0; // v
      for (int i = 0; i "lt" n; ++i)
      {
        double predY = model.Predict(dataX[i]);
        sumTop +=
          (dataY[i] - predY) * (dataY[i] - predY);
        sumBot +=
          (dataY[i] - meanY) * (dataY[i] - meanY);
      }
      return 1.0 - (sumTop / sumBot);
    }

    // ------------------------------------------------------

    public static double RootMSE(dynamic model,
      double[][] dataX, double[] dataY)
    {
      int n = dataX.Length;
      double sum = 0.0;
      for (int i = 0; i "lt" n; ++i)
      {
        double actualY = dataY[i];
        double predY = model.Predict(dataX[i]);
        sum += (actualY - predY) * (actualY - predY);
      }
      double mse = sum / n;
      return Math.Sqrt(mse);
    }

    // ------------------------------------------------------

    public static double BaselineAccuracy(double[][] dataX,
      double[] dataY, double pctClose)
    {
      // compute avg of y values
      int n = dataY.Length;
      double sum = 0.0;

      for (int i = 0; i "lt" n; ++i)
        sum += dataY[i];
      double meanY = sum / n;

      // always predict the mean
      int numCorrect = 0; int numWrong = 0;
      for (int i = 0; i "lt" dataX.Length; ++i)
      {
        double actualY = dataY[i];
        double predY = meanY;
        if (Math.Abs(predY - actualY) "lt"
          Math.Abs(pctClose * actualY))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / (numWrong + numCorrect);
    }

    // ------------------------------------------------------

    public static double BaselineMSE(double[][] dataX,
      double[] dataY)
    {
      // compute avg of y values
      int n = dataY.Length;
      double sum = 0.0;

      for (int i = 0; i "lt" n; ++i)
        sum += dataY[i];
      double meanY = sum / n;

      // always predict the mean
      sum = 0.0;
      for (int i = 0; i "lt" n; ++i)
      {
        double actualY = dataY[i];
        double predY = meanY;
        sum += (actualY - predY) * (actualY - predY);
      }
      return sum / n;
    }

    // ------------------------------------------------------

  } // class Eval

} // ns

Trainining data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning | Leave a comment

“AdaBoost.R2 Regression Using C#” in Visual Studio Magazine

Posted on June 11, 2026 by jamesdmccaffrey

I wrote an article titled “AdaBoost.R2 Regression Using C#” in the June 2026 edition of Microsoft Visual Studio Magazine. See https://visualstudiomagazine.com/articles/2026/06/01/adaboost-r2-regression-using-csharp.aspx.

The goal of a machine learning regression problem is to predict a single numeric value. For example, a bank might want to predict the maximum safe loan amount for a customer based on age, account balance, current debt, and so on. Regression techniques fall into two categories: those based on decision tree structures, and those that use non-tree techniques.

In a nutshell:

* AdaBoost.R2 regression is a machine learning technique used to predict a single numeric value
* AdaBoost.R2 builds a sequence of decision tree regressors where each accepted tree improves prediction compared to earlier trees.
* Final AdaBoost.R2 predictions are produced as a weighted median across the decision tree predictions.

Examples of non-tree techniques include linear regression, quadratic regression, kernel ridge regression, nearest neighbors regression, and neural network regression. The three most common tree-based regression techniques are random forest regression (which includes a variant called bagging tree regression), AdaBoost.R2 regression, and gradient boost regression. Each regression technique has pros and cons. If there was one regression technique that worked best for all types of datasets, there would only be one technique, not many.

My article presents a complete demo of AdaBoost.R2 implemented using the C# language. AdaBoost.R2 regression sequentially creates an ensemble (collection) of simple decision trees, where each tree is a bit better at prediction than the previous tree. For a given input x, the predicted y value is the weighted median of the predictions of the collection of decision trees.

The output of the article demo program is:

Begin AdaBoost.R2 (tree) regression from scratch C#

Loading synthetic train (200) and test (40) data

First three train X:
 -0.1660  0.4406 -0.9998 -0.3953 -0.7065
  0.0776 -0.1616  0.3704 -0.5911  0.7562
 -0.9452  0.3409 -0.1654  0.1174 -0.7192

First three train y:
  0.4840
  0.1568
  0.8054

Setting maxLearners = 100
Setting tree maxDepth = 5
Setting tree minSamples = 2
Setting tree minLeaf = 1

Training AdaBoost.R2 model
Done
Created 100 learners

Accuracy train (within 0.10): 0.8250
Accuracy test (within 0.10): 0.5250

MSE train: 0.0004
MSE test: 0.0022

Predicting for x =
 -0.1660  0.4406 -0.9998 -0.3953 -0.7065
Predicted y = 0.4890

End demo

I based my demo program directly on the 1997 source research paper, “Improving Regressors Using Boosting Techniques” (1997), by H. Drucker.

A weight value is assigned to each training data item, where the weight values sum to 1. Initially all training item weights are the same. For the demo data, because there are 200 items, each initial data item weight is 1/200 = 0.005.

Each training data item is fed to the current decision tree and an average loss/error for the current tree is computed. The average loss for the current tree is used to compute a beta value for the tree. A small value for beta means high confidence in the current tree, and vice versa.

The weights associated with each training data item are updated using the beta value for the current tree and the loss/error for each data item. At this point, a new set of training data items is probabilistically created from the source data items, where data items with high error have a greater probability of being selected than data items with small error.

The new training data items are fed to the next decision tree. Because the new training dataset has high-error items, the new tree will concentrate on learning to predict those difficult-to-predict items.

Each new training dataset is constructed probabilistically, and so there is a chance that a newly created decision tree is not better than the previous decision tree. In this case, the newly created decision tree is not added to the collection of trees. This is why the actual number of trees created may be less than the maxLearners parameter.

AdaBoost.R2 regression is not used nearly as often as random forest regression, or gradient boost regression. My AdaBoost.R2 demo is therefore most likely to be used with legacy systems that already use AdaBoost.R2 regression.

Like most of my friends, I get obsessed with all kinds of strange things. I have been obsessed with machine learning for many years.

For reasons that are unknown to me, I am moderately obsessed with movies that feature a creature that makes noises that are translated to “chittering” in the closed caption text. The are an amazing number of chittering instances in movies, but you’ll usually only notice them if you’re specifically looking for them.

Left: In the fantasy movie “The Dark Crystal” (192), Gelfling Jen and his girlfriend Kira, must overthrow the evil, ruling Skeksis by restoring a powerful broken Crystal. Early in the movie, Jen travels throw a creepy swamp with a chittering creature.

Right: In the fantasy movie “The Legend of Ochi” (2025), farm girl Yuri discovers a wounded baby primate-like creature and works to return it to its family. She is menaced by a hunting party led by her father and her stepbrother. When Yuri goes through a cave system, some insects are chittering while a bat watches.

Posted in Machine Learning | Leave a comment

An Example of RANSAC Regression Using Scikit on the Diabetes Dataset

Posted on June 10, 2026 by jamesdmccaffrey

The RANSAC (random sample consensus) regression technique can be used when the training dataset is known to have outliers (typically, blatant errors). RANSAC is a meta-heuristic that can be applied to any base regression technique, but is usually applied to basic linear regression.

The idea is:

inliers = []
loop many times
  take a very small sample from train data
  train a model
  use model on train data to find inliers
  if enough inliers
    train a "good" model on inliers
  end-if
end-loop
return best model found

I usually think of RANSAC as a kind of automatic outlier detection and removal, combined with regression. Before I go any further, let me note that the original RANSAC idea was developed in 1980 for a very specific image processing problem scenario, and the idea works quite well. But RANSAC has been applied to more general regression scenarios, and I am not a fan of RANSAC in those general scenarios.

I put together a RANSAC regression demo using the scikit-learn RANSACRegressor module on the Diabetes Dataset. The output of the demo:

Begin scikit RANSAC (random sample consensus)
 on Diabetes Dataset demo

Loading diabetes train (342), test (100) data
Done

First three X predictors:
[0.5900 1.0000 0.3210 0.1010 0.1570 0.0932 0.3800 0.4000
 0.4860 0.0870]
[0.4800 0.0000 0.2160 0.0870 0.1830 0.1032 0.7000 0.3000
 0.3892 0.0690]
[0.7200 1.0000 0.3050 0.0930 0.1560 0.0936 0.4100 0.4000
 0.4673 0.0850]

First three y targets:
0.1510
0.0750
0.1410

Creating scikit RANSAC model
Done

Training RANSAC model
Done.

Evaluating model

Accuracy (within 0.10) train = 0.1462
Accuracy (within 0.10) test = 0.1413

MSE train = 0.0040
MSE test = 0.0049

===============================

Creating, training base linear regression model

LR accuracy (within 0.10) train = 0.1813
LR accuracy (within 0.10) test = 0.2500

LR MSE train = 0.0029
LR MSE test = 0.0028

End demo

The RANSAC results were poor, as expected, and the RANSAC results were even worse than just plain linear regression, as expected.

The raw Diabetes Dataset looks like:

59, 2, 32.1, 101.00, 157,  93.2, 38, 4.00, 4.8598, 87, 151
48, 1, 21.6,  87.00, 183, 103.2, 70, 3.00, 3.8918, 69,  75
72, 2, 30.5,  93.00, 156,  93.6, 41, 4.00, 4.6728, 85, 141
. . .

Each line represents a patient. The first 10 values on each line are predictors. The last value on each line is the target value (a diabetes metric) to predict. The predictors are: age, sex, body mass index, blood pressure, serum cholesterol, low-density lipoproteins, high-density lipoproteins, total cholesterol, triglycerides, blood sugar. There are 442 data items.

The sex encoding isn’t explained anywhere but I suspect male = 1, female = 2 because there are 235 1 values and 206 2 values).

Note that this Diabetes Dataset, which is included as an example dataset in the Python language scikit-learn library, is not the same as the Pima Diabetes Dataset from the UCI dataset repository. See https://jamesmccaffreyblog.com/2026/02/03/the-origin-and-history-of-scikit-learn-diabetes-dataset/.

I tracked down, and read, the original 1980 RANSAC paper.

I converted the sex values from 1,2 into 0,1. Then I applied divide-by-constant normalization by dividing the 10 predictor columns by (100, 1, 100, 1000, 1000, 1000, 100, 10, 10, 1000) and the target y values by 1000. The resulting encoded and normalized data looks like:

0.5900, 1.0000, 0.3210, . . . 0.1510
0.4800, 0.0000, 0.2160, . . . 0.0750
0.7200, 1.0000, 0.3050, . . . 0.1410
. . .

I split the 442-items into a 342-item training set and a 100-item test set. I have done many experiments with the Diabetes Dataset and I’ve concluded the the default target value in the last column (a patient diabetes score) simply cannot be predicted well. But the variables in columns [4], [5], [6], [7], and [8] can be meaningfully predicted from the other columns.

In some sense, machine learning regression can be thought of as searching for hidden patterns in data.

Every cover of Playboy Magazine, except for the very first one (December 1953), has the company bunny logo somewhere. On most covers, the logo is clear and easy to see. But on some covers, the logo cleverly hidden. Looking for hidden items in a photo has an odd appeal to most people.

Left: On the cover of the issue from February 1993, the bunny logo is very difficult to spot. It’s incorporated into the long red threads on the model’s clothing. I added a white circle for you. Right: One of the most difficult hidden logos ever: on the cover of the July 1993 issue, the bunny logo is disguised as part of the water reflections. I circled it in red for you.

Demo program. Replace “lt” (less than) in the accuracy() function with the Boolean less-than operator symbol. (My blog editor chokes on symbols).

# diabetes_scikit_ransac.py
# RANSAC (random sample consensus)
# for the Diabetes Dataset

import numpy as np
from sklearn.linear_model import RANSACRegressor

# RANSACRegressor(estimator=None, *, min_samples=None,
# residual_threshold=None, is_data_valid=None,
# is_model_valid=None, max_trials=100, max_skips=inf,
# stop_n_inliers=inf, stop_score=inf,
# stop_probability=0.99, loss='absolute_error',
# random_state=None)

# -----------------------------------------------------------

np.set_printoptions(precision=4, suppress=True,
  floatmode='fixed', linewidth=60)

# -----------------------------------------------------------

def accuracy(model, data_X, data_y, pct_close):
  n = len(data_X)
  n_correct = 0; n_wrong = 0
  for i in range(n):
    x = data_X[i].reshape(1,-1)
    y = data_y[i]
    y_pred = model.predict(x)[0]

    if np.abs(y - y_pred) "lt" np.abs(y * pct_close):
      n_correct += 1
    else: 
      n_wrong += 1
  return n_correct / (n_correct + n_wrong)

def mse(model, data_X, data_y):
  n = len(data_X)
  sum = 0.0
  for i in range(n):
    actual_y = data_y[i]
    pred_y = model.predict(data_X[i].reshape(1, -1))[0]
    diff = actual_y - pred_y
    sum += diff * diff
  return sum /n

# -----------------------------------------------------------

print("\nBegin scikit RANSAC (random sample consensus) " +\
 "on Diabetes Dataset demo ")

print("\nLoading diabetes train (342), test (100) data ")
train_file = ".\\Data\\diabetes_norm_train_342.txt"

cols_X = [0,1,2,3,4,5,6,7,8,9]  
col_y = 10  # cols # 4 5 6 7 8 are much better
train_X = np.loadtxt(train_file, comments="#",
  usecols=cols_X, delimiter=",",  dtype=np.float64)
train_y = np.loadtxt(train_file, comments="#",
  usecols=col_y, delimiter=",",  dtype=np.float64)

test_file = ".\\Data\\diabetes_norm_test_100.txt"
test_X = np.loadtxt(test_file, comments="#",
  usecols=cols_X, delimiter=",",  dtype=np.float64)
test_y = np.loadtxt(test_file, comments="#",
  usecols=col_y, delimiter=",",  dtype=np.float64)
print("Done ")

# alternative normalization and split
# from sklearn.datasets import load_diabetes
# from sklearn.model_selection import train_test_split
# X, y = load_diabetes(return_X_y=True, scaled=True)
# train_X, test_X, train_y, test_y = \
#   train_test_split(X, y, random_state=0)  # 25% test

print("\nFirst three X predictors: ")
for i in range(3):
  print(train_X[i])
print("\nFirst three y targets: ")
for i in range(3):
  print("%0.4f" % train_y[i])

print("\nCreating scikit RANSAC model ")
model = RANSACRegressor(random_state=0)
print("Done ")

print("\nTraining RANSAC model ")
model.fit(train_X, train_y)
print("Done. ")

print("\nEvaluating model ")
acc_train = accuracy(model, train_X, train_y, 0.10)
acc_test = accuracy(model, test_X, test_y, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
print("Accuracy (within 0.10) test = %0.4f " % \
  acc_test)

mse_train = mse(model, train_X, train_y)
mse_test = mse(model, test_X, test_y)
print("\nMSE train = %0.4f " % mse_train)
print("MSE test = %0.4f " % mse_test)

print("\n=============================== ")

print("\nCreating, training base linear regression model")
from sklearn.linear_model import LinearRegression
lr_model = LinearRegression()
lr_model.fit(train_X, train_y)

acc_train = accuracy(lr_model, train_X, train_y, 0.10)
acc_test = accuracy(lr_model, test_X, test_y, 0.10)
print("\nLR accuracy (within 0.10) train = %0.4f " % \
  acc_train)
print("LR accuracy (within 0.10) test = %0.4f " % \
  acc_test)

mse_train = mse(lr_model, train_X, train_y)
mse_test = mse(lr_model, test_X, test_y)
print("\nLR MSE train = %0.4f " % mse_train)
print("LR MSE test = %0.4f " % mse_test)

print("\nEnd demo ")

Training data:


# diabetes_norm_train_342.txt
# cols [0] to [9] predictors. col [10] target
# norm division constants:
# 100, -1, 100, 1000, 1000, 1000, 100, 10, 10, 1000, 1000
#
0.5900, 1.0000, 0.3210, 0.1010, 0.1570, 0.0932, 0.3800, 0.4000, 0.4860, 0.0870, 0.1510
0.4800, 0.0000, 0.2160, 0.0870, 0.1830, 0.1032, 0.7000, 0.3000, 0.3892, 0.0690, 0.0750
0.7200, 1.0000, 0.3050, 0.0930, 0.1560, 0.0936, 0.4100, 0.4000, 0.4673, 0.0850, 0.1410
0.2400, 0.0000, 0.2530, 0.0840, 0.1980, 0.1314, 0.4000, 0.5000, 0.4890, 0.0890, 0.2060
0.5000, 0.0000, 0.2300, 0.1010, 0.1920, 0.1254, 0.5200, 0.4000, 0.4291, 0.0800, 0.1350
0.2300, 0.0000, 0.2260, 0.0890, 0.1390, 0.0648, 0.6100, 0.2000, 0.4190, 0.0680, 0.0970
0.3600, 1.0000, 0.2200, 0.0900, 0.1600, 0.0996, 0.5000, 0.3000, 0.3951, 0.0820, 0.1380
0.6600, 1.0000, 0.2620, 0.1140, 0.2550, 0.1850, 0.5600, 0.4550, 0.4249, 0.0920, 0.0630
0.6000, 1.0000, 0.3210, 0.0830, 0.1790, 0.1194, 0.4200, 0.4000, 0.4477, 0.0940, 0.1100
0.2900, 0.0000, 0.3000, 0.0850, 0.1800, 0.0934, 0.4300, 0.4000, 0.5385, 0.0880, 0.3100
0.2200, 0.0000, 0.1860, 0.0970, 0.1140, 0.0576, 0.4600, 0.2000, 0.3951, 0.0830, 0.1010
0.5600, 1.0000, 0.2800, 0.0850, 0.1840, 0.1448, 0.3200, 0.6000, 0.3584, 0.0770, 0.0690
0.5300, 0.0000, 0.2370, 0.0920, 0.1860, 0.1092, 0.6200, 0.3000, 0.4304, 0.0810, 0.1790
0.5000, 1.0000, 0.2620, 0.0970, 0.1860, 0.1054, 0.4900, 0.4000, 0.5063, 0.0880, 0.1850
0.6100, 0.0000, 0.2400, 0.0910, 0.2020, 0.1154, 0.7200, 0.3000, 0.4291, 0.0730, 0.1180
0.3400, 1.0000, 0.2470, 0.1180, 0.2540, 0.1842, 0.3900, 0.7000, 0.5037, 0.0810, 0.1710
0.4700, 0.0000, 0.3030, 0.1090, 0.2070, 0.1002, 0.7000, 0.3000, 0.5215, 0.0980, 0.1660
0.6800, 1.0000, 0.2750, 0.1110, 0.2140, 0.1470, 0.3900, 0.5000, 0.4942, 0.0910, 0.1440
0.3800, 0.0000, 0.2540, 0.0840, 0.1620, 0.1030, 0.4200, 0.4000, 0.4443, 0.0870, 0.0970
0.4100, 0.0000, 0.2470, 0.0830, 0.1870, 0.1082, 0.6000, 0.3000, 0.4543, 0.0780, 0.1680
0.3500, 0.0000, 0.2110, 0.0820, 0.1560, 0.0878, 0.5000, 0.3000, 0.4511, 0.0950, 0.0680
0.2500, 1.0000, 0.2430, 0.0950, 0.1620, 0.0986, 0.5400, 0.3000, 0.3850, 0.0870, 0.0490
0.2500, 0.0000, 0.2600, 0.0920, 0.1870, 0.1204, 0.5600, 0.3000, 0.3970, 0.0880, 0.0680
0.6100, 1.0000, 0.3200, 0.1037, 0.2100, 0.0852, 0.3500, 0.6000, 0.6107, 0.1240, 0.2450
0.3100, 0.0000, 0.2970, 0.0880, 0.1670, 0.1034, 0.4800, 0.4000, 0.4357, 0.0780, 0.1840
0.3000, 1.0000, 0.2520, 0.0830, 0.1780, 0.1184, 0.3400, 0.5000, 0.4852, 0.0830, 0.2020
0.1900, 0.0000, 0.1920, 0.0870, 0.1240, 0.0540, 0.5700, 0.2000, 0.4174, 0.0900, 0.1370
0.4200, 0.0000, 0.3190, 0.0830, 0.1580, 0.0876, 0.5300, 0.3000, 0.4466, 0.1010, 0.0850
0.6300, 0.0000, 0.2440, 0.0730, 0.1600, 0.0914, 0.4800, 0.3000, 0.4635, 0.0780, 0.1310
0.6700, 1.0000, 0.2580, 0.1130, 0.1580, 0.0542, 0.6400, 0.2000, 0.5293, 0.1040, 0.2830
0.3200, 0.0000, 0.3050, 0.0890, 0.1820, 0.1106, 0.5600, 0.3000, 0.4344, 0.0890, 0.1290
0.4200, 0.0000, 0.2030, 0.0710, 0.1610, 0.0812, 0.6600, 0.2000, 0.4234, 0.0810, 0.0590
0.5800, 1.0000, 0.3800, 0.1030, 0.1500, 0.1072, 0.2200, 0.7000, 0.4644, 0.0980, 0.3410
0.5700, 0.0000, 0.2170, 0.0940, 0.1570, 0.0580, 0.8200, 0.2000, 0.4443, 0.0920, 0.0870
0.5300, 0.0000, 0.2050, 0.0780, 0.1470, 0.0842, 0.5200, 0.3000, 0.3989, 0.0750, 0.0650
0.6200, 1.0000, 0.2350, 0.0803, 0.2250, 0.1128, 0.8600, 0.2620, 0.4875, 0.0960, 0.1020
0.5200, 0.0000, 0.2850, 0.1100, 0.1950, 0.0972, 0.6000, 0.3000, 0.5242, 0.0850, 0.2650
0.4600, 0.0000, 0.2740, 0.0780, 0.1710, 0.0880, 0.5800, 0.3000, 0.4828, 0.0900, 0.2760
0.4800, 1.0000, 0.3300, 0.1230, 0.2530, 0.1636, 0.4400, 0.6000, 0.5425, 0.0970, 0.2520
0.4800, 1.0000, 0.2770, 0.0730, 0.1910, 0.1194, 0.4600, 0.4000, 0.4852, 0.0920, 0.0900
0.5000, 1.0000, 0.2560, 0.1010, 0.2290, 0.1622, 0.4300, 0.5000, 0.4779, 0.1140, 0.1000
0.2100, 0.0000, 0.2010, 0.0630, 0.1350, 0.0690, 0.5400, 0.3000, 0.4094, 0.0890, 0.0550
0.3200, 1.0000, 0.2540, 0.0903, 0.1530, 0.1004, 0.3400, 0.4500, 0.4533, 0.0830, 0.0610
0.5400, 0.0000, 0.2420, 0.0740, 0.2040, 0.1090, 0.8200, 0.2000, 0.4174, 0.1090, 0.0920
0.6100, 1.0000, 0.3270, 0.0970, 0.1770, 0.1184, 0.2900, 0.6000, 0.4997, 0.0870, 0.2590
0.5600, 1.0000, 0.2310, 0.1040, 0.1810, 0.1164, 0.4700, 0.4000, 0.4477, 0.0790, 0.0530
0.3300, 0.0000, 0.2530, 0.0850, 0.1550, 0.0850, 0.5100, 0.3000, 0.4554, 0.0700, 0.1900
0.2700, 0.0000, 0.1960, 0.0780, 0.1280, 0.0680, 0.4300, 0.3000, 0.4443, 0.0710, 0.1420
0.6700, 1.0000, 0.2250, 0.0980, 0.1910, 0.1192, 0.6100, 0.3000, 0.3989, 0.0860, 0.0750
0.3700, 1.0000, 0.2770, 0.0930, 0.1800, 0.1194, 0.3000, 0.6000, 0.5030, 0.0880, 0.1420
0.5800, 0.0000, 0.2570, 0.0990, 0.1570, 0.0916, 0.4900, 0.3000, 0.4407, 0.0930, 0.1550
0.6500, 1.0000, 0.2790, 0.1030, 0.1590, 0.0968, 0.4200, 0.4000, 0.4615, 0.0860, 0.2250
0.3400, 0.0000, 0.2550, 0.0930, 0.2180, 0.1440, 0.5700, 0.4000, 0.4443, 0.0880, 0.0590
0.4600, 0.0000, 0.2490, 0.1150, 0.1980, 0.1296, 0.5400, 0.4000, 0.4277, 0.1030, 0.1040
0.3500, 0.0000, 0.2870, 0.0970, 0.2040, 0.1268, 0.6400, 0.3000, 0.4190, 0.0930, 0.1820
0.3700, 0.0000, 0.2180, 0.0840, 0.1840, 0.1010, 0.7300, 0.3000, 0.3912, 0.0930, 0.1280
0.3700, 0.0000, 0.3020, 0.0870, 0.1660, 0.0960, 0.4000, 0.4150, 0.5011, 0.0870, 0.0520
0.4100, 0.0000, 0.2050, 0.0800, 0.1240, 0.0488, 0.6400, 0.2000, 0.4025, 0.0750, 0.0370
0.6000, 0.0000, 0.2040, 0.1050, 0.1980, 0.0784, 0.9900, 0.2000, 0.4635, 0.0790, 0.1700
0.6600, 1.0000, 0.2400, 0.0980, 0.2360, 0.1464, 0.5800, 0.4000, 0.5063, 0.0960, 0.1700
0.2900, 0.0000, 0.2600, 0.0830, 0.1410, 0.0652, 0.6400, 0.2000, 0.4078, 0.0830, 0.0610
0.3700, 1.0000, 0.2680, 0.0790, 0.1570, 0.0980, 0.2800, 0.6000, 0.5043, 0.0960, 0.1440
0.4100, 1.0000, 0.2570, 0.0830, 0.1810, 0.1066, 0.6600, 0.3000, 0.3738, 0.0850, 0.0520
0.3900, 0.0000, 0.2290, 0.0770, 0.2040, 0.1432, 0.4600, 0.4000, 0.4304, 0.0740, 0.1280
0.6700, 1.0000, 0.2400, 0.0830, 0.1430, 0.0772, 0.4900, 0.3000, 0.4431, 0.0940, 0.0710
0.3600, 1.0000, 0.2410, 0.1120, 0.1930, 0.1250, 0.3500, 0.6000, 0.5106, 0.0950, 0.1630
0.4600, 1.0000, 0.2470, 0.0850, 0.1740, 0.1232, 0.3000, 0.6000, 0.4644, 0.0960, 0.1500
0.6000, 1.0000, 0.2500, 0.0897, 0.1850, 0.1208, 0.4600, 0.4020, 0.4511, 0.0920, 0.0970
0.5900, 1.0000, 0.2360, 0.0830, 0.1650, 0.1000, 0.4700, 0.4000, 0.4500, 0.0920, 0.1600
0.5300, 0.0000, 0.2210, 0.0930, 0.1340, 0.0762, 0.4600, 0.3000, 0.4078, 0.0960, 0.1780
0.4800, 0.0000, 0.1990, 0.0910, 0.1890, 0.1096, 0.6900, 0.3000, 0.3951, 0.1010, 0.0480
0.4800, 0.0000, 0.2950, 0.1310, 0.2070, 0.1322, 0.4700, 0.4000, 0.4935, 0.1060, 0.2700
0.6600, 1.0000, 0.2600, 0.0910, 0.2640, 0.1466, 0.6500, 0.4000, 0.5568, 0.0870, 0.2020
0.5200, 1.0000, 0.2450, 0.0940, 0.2170, 0.1494, 0.4800, 0.5000, 0.4585, 0.0890, 0.1110
0.5200, 1.0000, 0.2660, 0.1110, 0.2090, 0.1264, 0.6100, 0.3000, 0.4682, 0.1090, 0.0850
0.4600, 1.0000, 0.2350, 0.0870, 0.1810, 0.1148, 0.4400, 0.4000, 0.4710, 0.0980, 0.0420
0.4000, 1.0000, 0.2900, 0.1150, 0.0970, 0.0472, 0.3500, 0.2770, 0.4304, 0.0950, 0.1700
0.2200, 0.0000, 0.2300, 0.0730, 0.1610, 0.0978, 0.5400, 0.3000, 0.3829, 0.0910, 0.2000
0.5000, 0.0000, 0.2100, 0.0880, 0.1400, 0.0718, 0.3500, 0.4000, 0.5112, 0.0710, 0.2520
0.2000, 0.0000, 0.2290, 0.0870, 0.1910, 0.1282, 0.5300, 0.4000, 0.3892, 0.0850, 0.1130
0.6800, 0.0000, 0.2750, 0.1070, 0.2410, 0.1496, 0.6400, 0.4000, 0.4920, 0.0900, 0.1430
0.5200, 1.0000, 0.2430, 0.0860, 0.1970, 0.1336, 0.4400, 0.5000, 0.4575, 0.0910, 0.0510
0.4400, 0.0000, 0.2310, 0.0870, 0.2130, 0.1264, 0.7700, 0.3000, 0.3871, 0.0720, 0.0520
0.3800, 0.0000, 0.2730, 0.0810, 0.1460, 0.0816, 0.4700, 0.3000, 0.4466, 0.0810, 0.2100
0.4900, 0.0000, 0.2270, 0.0653, 0.1680, 0.0962, 0.6200, 0.2710, 0.3892, 0.0600, 0.0650
0.6100, 0.0000, 0.3300, 0.0950, 0.1820, 0.1148, 0.5400, 0.3000, 0.4190, 0.0740, 0.1410
0.2900, 1.0000, 0.1940, 0.0830, 0.1520, 0.1058, 0.3900, 0.4000, 0.3584, 0.0830, 0.0550
0.6100, 0.0000, 0.2580, 0.0980, 0.2350, 0.1258, 0.7600, 0.3000, 0.5112, 0.0820, 0.1340
0.3400, 1.0000, 0.2260, 0.0750, 0.1660, 0.0918, 0.6000, 0.3000, 0.4263, 0.1080, 0.0420
0.3600, 0.0000, 0.2190, 0.0890, 0.1890, 0.1052, 0.6800, 0.3000, 0.4369, 0.0960, 0.1110
0.5200, 0.0000, 0.2400, 0.0830, 0.1670, 0.0866, 0.7100, 0.2000, 0.3850, 0.0940, 0.0980
0.6100, 0.0000, 0.3120, 0.0790, 0.2350, 0.1568, 0.4700, 0.5000, 0.5050, 0.0960, 0.1640
0.4300, 0.0000, 0.2680, 0.1230, 0.1930, 0.1022, 0.6700, 0.3000, 0.4779, 0.0940, 0.0480
0.3500, 0.0000, 0.2040, 0.0650, 0.1870, 0.1056, 0.6700, 0.2790, 0.4277, 0.0780, 0.0960
0.2700, 0.0000, 0.2480, 0.0910, 0.1890, 0.1068, 0.6900, 0.3000, 0.4190, 0.0690, 0.0900
0.2900, 0.0000, 0.2100, 0.0710, 0.1560, 0.0970, 0.3800, 0.4000, 0.4654, 0.0900, 0.1620
0.6400, 1.0000, 0.2730, 0.1090, 0.1860, 0.1076, 0.3800, 0.5000, 0.5308, 0.0990, 0.1500
0.4100, 0.0000, 0.3460, 0.0873, 0.2050, 0.1426, 0.4100, 0.5000, 0.4673, 0.1100, 0.2790
0.4900, 1.0000, 0.2590, 0.0910, 0.1780, 0.1066, 0.5200, 0.3000, 0.4575, 0.0750, 0.0920
0.4800, 0.0000, 0.2040, 0.0980, 0.2090, 0.1394, 0.4600, 0.5000, 0.4771, 0.0780, 0.0830
0.5300, 0.0000, 0.2800, 0.0880, 0.2330, 0.1438, 0.5800, 0.4000, 0.5050, 0.0910, 0.1280
0.5300, 1.0000, 0.2220, 0.1130, 0.1970, 0.1152, 0.6700, 0.3000, 0.4304, 0.1000, 0.1020
0.2300, 0.0000, 0.2900, 0.0900, 0.2160, 0.1314, 0.6500, 0.3000, 0.4585, 0.0910, 0.3020
0.6500, 1.0000, 0.3020, 0.0980, 0.2190, 0.1606, 0.4000, 0.5000, 0.4522, 0.0840, 0.1980
0.4100, 0.0000, 0.3240, 0.0940, 0.1710, 0.1044, 0.5600, 0.3000, 0.3970, 0.0760, 0.0950
0.5500, 1.0000, 0.2340, 0.0830, 0.1660, 0.1016, 0.4600, 0.4000, 0.4522, 0.0960, 0.0530
0.2200, 0.0000, 0.1930, 0.0820, 0.1560, 0.0932, 0.5200, 0.3000, 0.3989, 0.0710, 0.1340
0.5600, 0.0000, 0.3100, 0.0787, 0.1870, 0.1414, 0.3400, 0.5500, 0.4060, 0.0900, 0.1440
0.5400, 1.0000, 0.3060, 0.1033, 0.1440, 0.0798, 0.3000, 0.4800, 0.5142, 0.1010, 0.2320
0.5900, 1.0000, 0.2550, 0.0953, 0.1900, 0.1394, 0.3500, 0.5430, 0.4357, 0.1170, 0.0810
0.6000, 1.0000, 0.2340, 0.0880, 0.1530, 0.0898, 0.5800, 0.3000, 0.3258, 0.0950, 0.1040
0.5400, 0.0000, 0.2680, 0.0870, 0.2060, 0.1220, 0.6800, 0.3000, 0.4382, 0.0800, 0.0590
0.2500, 0.0000, 0.2830, 0.0870, 0.1930, 0.1280, 0.4900, 0.4000, 0.4382, 0.0920, 0.2460
0.5400, 1.0000, 0.2770, 0.1130, 0.2000, 0.1284, 0.3700, 0.5000, 0.5153, 0.1130, 0.2970
0.5500, 0.0000, 0.3660, 0.1130, 0.1990, 0.0944, 0.4300, 0.4630, 0.5730, 0.0970, 0.2580
0.4000, 1.0000, 0.2650, 0.0930, 0.2360, 0.1470, 0.3700, 0.7000, 0.5561, 0.0920, 0.2290
0.6200, 1.0000, 0.3180, 0.1150, 0.1990, 0.1286, 0.4400, 0.5000, 0.4883, 0.0980, 0.2750
0.6500, 0.0000, 0.2440, 0.1200, 0.2220, 0.1356, 0.3700, 0.6000, 0.5509, 0.1240, 0.2810
0.3300, 1.0000, 0.2540, 0.1020, 0.2060, 0.1410, 0.3900, 0.5000, 0.4868, 0.1050, 0.1790
0.5300, 0.0000, 0.2200, 0.0940, 0.1750, 0.0880, 0.5900, 0.3000, 0.4942, 0.0980, 0.2000
0.3500, 0.0000, 0.2680, 0.0980, 0.1620, 0.1036, 0.4500, 0.4000, 0.4205, 0.0860, 0.2000
0.6600, 0.0000, 0.2800, 0.1010, 0.1950, 0.1292, 0.4000, 0.5000, 0.4860, 0.0940, 0.1730
0.6200, 1.0000, 0.3390, 0.1010, 0.2210, 0.1564, 0.3500, 0.6000, 0.4997, 0.1030, 0.1800
0.5000, 1.0000, 0.2960, 0.0943, 0.3000, 0.2424, 0.3300, 0.9090, 0.4812, 0.1090, 0.0840
0.4700, 0.0000, 0.2860, 0.0970, 0.1640, 0.0906, 0.5600, 0.3000, 0.4466, 0.0880, 0.1210
0.4700, 1.0000, 0.2560, 0.0940, 0.1650, 0.0748, 0.4000, 0.4000, 0.5526, 0.0930, 0.1610
0.2400, 0.0000, 0.2070, 0.0870, 0.1490, 0.0806, 0.6100, 0.2000, 0.3611, 0.0780, 0.0990
0.5800, 1.0000, 0.2620, 0.0910, 0.2170, 0.1242, 0.7100, 0.3000, 0.4691, 0.0680, 0.1090
0.3400, 0.0000, 0.2060, 0.0870, 0.1850, 0.1122, 0.5800, 0.3000, 0.4304, 0.0740, 0.1150
0.5100, 0.0000, 0.2790, 0.0960, 0.1960, 0.1222, 0.4200, 0.5000, 0.5069, 0.1200, 0.2680
0.3100, 1.0000, 0.3530, 0.1250, 0.1870, 0.1124, 0.4800, 0.4000, 0.4890, 0.1090, 0.2740
0.2200, 0.0000, 0.1990, 0.0750, 0.1750, 0.1086, 0.5400, 0.3000, 0.4127, 0.0720, 0.1580
0.5300, 1.0000, 0.2440, 0.0920, 0.2140, 0.1460, 0.5000, 0.4000, 0.4500, 0.0970, 0.1070
0.3700, 1.0000, 0.2140, 0.0830, 0.1280, 0.0696, 0.4900, 0.3000, 0.3850, 0.0840, 0.0830
0.2800, 0.0000, 0.3040, 0.0850, 0.1980, 0.1156, 0.6700, 0.3000, 0.4344, 0.0800, 0.1030
0.4700, 0.0000, 0.3160, 0.0840, 0.1540, 0.0880, 0.3000, 0.5100, 0.5199, 0.1050, 0.2720
0.2300, 0.0000, 0.1880, 0.0780, 0.1450, 0.0720, 0.6300, 0.2000, 0.3912, 0.0860, 0.0850
0.5000, 0.0000, 0.3100, 0.1230, 0.1780, 0.1050, 0.4800, 0.4000, 0.4828, 0.0880, 0.2800
0.5800, 1.0000, 0.3670, 0.1170, 0.1660, 0.0938, 0.4400, 0.4000, 0.4949, 0.1090, 0.3360
0.5500, 0.0000, 0.3210, 0.1100, 0.1640, 0.0842, 0.4200, 0.4000, 0.5242, 0.0900, 0.2810
0.6000, 1.0000, 0.2770, 0.1070, 0.1670, 0.1146, 0.3800, 0.4000, 0.4277, 0.0950, 0.1180
0.4100, 0.0000, 0.3080, 0.0810, 0.2140, 0.1520, 0.2800, 0.7600, 0.5136, 0.1230, 0.3170
0.6000, 1.0000, 0.2750, 0.1060, 0.2290, 0.1438, 0.5100, 0.4000, 0.5142, 0.0910, 0.2350
0.4000, 0.0000, 0.2690, 0.0920, 0.2030, 0.1198, 0.7000, 0.3000, 0.4190, 0.0810, 0.0600
0.5700, 1.0000, 0.3070, 0.0900, 0.2040, 0.1478, 0.3400, 0.6000, 0.4710, 0.0930, 0.1740
0.3700, 0.0000, 0.3830, 0.1130, 0.1650, 0.0946, 0.5300, 0.3000, 0.4466, 0.0790, 0.2590
0.4000, 1.0000, 0.3190, 0.0950, 0.1980, 0.1356, 0.3800, 0.5000, 0.4804, 0.0930, 0.1780
0.3300, 0.0000, 0.3500, 0.0890, 0.2000, 0.1304, 0.4200, 0.4760, 0.4927, 0.1010, 0.1280
0.3200, 1.0000, 0.2780, 0.0890, 0.2160, 0.1462, 0.5500, 0.4000, 0.4304, 0.0910, 0.0960
0.3500, 1.0000, 0.2590, 0.0810, 0.1740, 0.1024, 0.3100, 0.6000, 0.5313, 0.0820, 0.1260
0.5500, 0.0000, 0.3290, 0.1020, 0.1640, 0.1062, 0.4100, 0.4000, 0.4431, 0.0890, 0.2880
0.4900, 0.0000, 0.2600, 0.0930, 0.1830, 0.1002, 0.6400, 0.3000, 0.4543, 0.0880, 0.0880
0.3900, 1.0000, 0.2630, 0.1150, 0.2180, 0.1582, 0.3200, 0.7000, 0.4935, 0.1090, 0.2920
0.6000, 1.0000, 0.2230, 0.1130, 0.1860, 0.1258, 0.4600, 0.4000, 0.4263, 0.0940, 0.0710
0.6700, 1.0000, 0.2830, 0.0930, 0.2040, 0.1322, 0.4900, 0.4000, 0.4736, 0.0920, 0.1970
0.4100, 1.0000, 0.3200, 0.1090, 0.2510, 0.1706, 0.4900, 0.5000, 0.5056, 0.1030, 0.1860
0.4400, 0.0000, 0.2540, 0.0950, 0.1620, 0.0926, 0.5300, 0.3000, 0.4407, 0.0830, 0.0250
0.4800, 1.0000, 0.2330, 0.0893, 0.2120, 0.1428, 0.4600, 0.4610, 0.4754, 0.0980, 0.0840
0.4500, 0.0000, 0.2030, 0.0743, 0.1900, 0.1262, 0.4900, 0.3880, 0.4304, 0.0790, 0.0960
0.4700, 0.0000, 0.3040, 0.1200, 0.1990, 0.1200, 0.4600, 0.4000, 0.5106, 0.0870, 0.1950
0.4600, 0.0000, 0.2060, 0.0730, 0.1720, 0.1070, 0.5100, 0.3000, 0.4249, 0.0800, 0.0530
0.3600, 1.0000, 0.3230, 0.1150, 0.2860, 0.1994, 0.3900, 0.7000, 0.5472, 0.1120, 0.2170
0.3400, 0.0000, 0.2920, 0.0730, 0.1720, 0.1082, 0.4900, 0.4000, 0.4304, 0.0910, 0.1720
0.5300, 1.0000, 0.3310, 0.1170, 0.1830, 0.1190, 0.4800, 0.4000, 0.4382, 0.1060, 0.1310
0.6100, 0.0000, 0.2460, 0.1010, 0.2090, 0.1068, 0.7700, 0.3000, 0.4836, 0.0880, 0.2140
0.3700, 0.0000, 0.2020, 0.0810, 0.1620, 0.0878, 0.6300, 0.3000, 0.4025, 0.0880, 0.0590
0.3300, 1.0000, 0.2080, 0.0840, 0.1250, 0.0702, 0.4600, 0.3000, 0.3784, 0.0660, 0.0700
0.6800, 0.0000, 0.3280, 0.1057, 0.2050, 0.1164, 0.4000, 0.5130, 0.5493, 0.1170, 0.2200
0.4900, 1.0000, 0.3190, 0.0940, 0.2340, 0.1558, 0.3400, 0.7000, 0.5398, 0.1220, 0.2680
0.4800, 0.0000, 0.2390, 0.1090, 0.2320, 0.1052, 0.3700, 0.6000, 0.6107, 0.0960, 0.1520
0.5500, 1.0000, 0.2450, 0.0840, 0.1790, 0.1058, 0.6600, 0.3000, 0.3584, 0.0870, 0.0470
0.4300, 0.0000, 0.2210, 0.0660, 0.1340, 0.0772, 0.4500, 0.3000, 0.4078, 0.0800, 0.0740
0.6000, 1.0000, 0.3300, 0.0970, 0.2170, 0.1256, 0.4500, 0.5000, 0.5447, 0.1120, 0.2950
0.3100, 1.0000, 0.1900, 0.0930, 0.1370, 0.0730, 0.4700, 0.3000, 0.4443, 0.0780, 0.1010
0.5300, 1.0000, 0.2730, 0.0820, 0.1190, 0.0550, 0.3900, 0.3000, 0.4828, 0.0930, 0.1510
0.6700, 0.0000, 0.2280, 0.0870, 0.1660, 0.0986, 0.5200, 0.3000, 0.4344, 0.0920, 0.1270
0.6100, 1.0000, 0.2820, 0.1060, 0.2040, 0.1320, 0.5200, 0.4000, 0.4605, 0.0960, 0.2370
0.6200, 0.0000, 0.2890, 0.0873, 0.2060, 0.1272, 0.3300, 0.6240, 0.5434, 0.0990, 0.2250
0.6000, 0.0000, 0.2560, 0.0870, 0.2070, 0.1258, 0.6900, 0.3000, 0.4111, 0.0840, 0.0810
0.4200, 0.0000, 0.2490, 0.0910, 0.2040, 0.1418, 0.3800, 0.5000, 0.4796, 0.0890, 0.1510
0.3800, 1.0000, 0.2680, 0.1050, 0.1810, 0.1192, 0.3700, 0.5000, 0.4820, 0.0910, 0.1070
0.6200, 0.0000, 0.2240, 0.0790, 0.2220, 0.1474, 0.5900, 0.4000, 0.4357, 0.0760, 0.0640
0.6100, 1.0000, 0.2690, 0.1110, 0.2360, 0.1724, 0.3900, 0.6000, 0.4812, 0.0890, 0.1380
0.6100, 1.0000, 0.2310, 0.1130, 0.1860, 0.1144, 0.4700, 0.4000, 0.4812, 0.1050, 0.1850
0.5300, 0.0000, 0.2860, 0.0880, 0.1710, 0.0988, 0.4100, 0.4000, 0.5050, 0.0990, 0.2650
0.2800, 1.0000, 0.2470, 0.0970, 0.1750, 0.0996, 0.3200, 0.5000, 0.5380, 0.0870, 0.1010
0.2600, 1.0000, 0.3030, 0.0890, 0.2180, 0.1522, 0.3100, 0.7000, 0.5159, 0.0820, 0.1370
0.3000, 0.0000, 0.2130, 0.0870, 0.1340, 0.0630, 0.6300, 0.2000, 0.3689, 0.0660, 0.1430
0.5000, 0.0000, 0.2610, 0.1090, 0.2430, 0.1606, 0.6200, 0.4000, 0.4625, 0.0890, 0.1410
0.4800, 0.0000, 0.2020, 0.0950, 0.1870, 0.1174, 0.5300, 0.4000, 0.4419, 0.0850, 0.0790
0.5100, 0.0000, 0.2520, 0.1030, 0.1760, 0.1122, 0.3700, 0.5000, 0.4898, 0.0900, 0.2920
0.4700, 1.0000, 0.2250, 0.0820, 0.1310, 0.0668, 0.4100, 0.3000, 0.4754, 0.0890, 0.1780
0.6400, 1.0000, 0.2350, 0.0970, 0.2030, 0.1290, 0.5900, 0.3000, 0.4318, 0.0770, 0.0910
0.5100, 1.0000, 0.2590, 0.0760, 0.2400, 0.1690, 0.3900, 0.6000, 0.5075, 0.0960, 0.1160
0.3000, 0.0000, 0.2090, 0.1040, 0.1520, 0.0838, 0.4700, 0.3000, 0.4663, 0.0970, 0.0860
0.5600, 1.0000, 0.2870, 0.0990, 0.2080, 0.1464, 0.3900, 0.5000, 0.4727, 0.0970, 0.1220
0.4200, 0.0000, 0.2210, 0.0850, 0.2130, 0.1386, 0.6000, 0.4000, 0.4277, 0.0940, 0.0720
0.6200, 1.0000, 0.2670, 0.1150, 0.1830, 0.1240, 0.3500, 0.5000, 0.4788, 0.1000, 0.1290
0.3400, 0.0000, 0.3140, 0.0870, 0.1490, 0.0938, 0.4600, 0.3000, 0.3829, 0.0770, 0.1420
0.6000, 0.0000, 0.2220, 0.1047, 0.2210, 0.1054, 0.6000, 0.3680, 0.5628, 0.0930, 0.0900
0.6400, 0.0000, 0.2100, 0.0923, 0.2270, 0.1468, 0.6500, 0.3490, 0.4331, 0.1020, 0.1580
0.3900, 1.0000, 0.2120, 0.0900, 0.1820, 0.1104, 0.6000, 0.3000, 0.4060, 0.0980, 0.0390
0.7100, 1.0000, 0.2650, 0.1050, 0.2810, 0.1736, 0.5500, 0.5000, 0.5568, 0.0840, 0.1960
0.4800, 1.0000, 0.2920, 0.1100, 0.2180, 0.1516, 0.3900, 0.6000, 0.4920, 0.0980, 0.2220
0.7900, 1.0000, 0.2700, 0.1030, 0.1690, 0.1108, 0.3700, 0.5000, 0.4663, 0.1100, 0.2770
0.4000, 0.0000, 0.3070, 0.0990, 0.1770, 0.0854, 0.5000, 0.4000, 0.5338, 0.0850, 0.0990
0.4900, 1.0000, 0.2880, 0.0920, 0.2070, 0.1400, 0.4400, 0.5000, 0.4745, 0.0920, 0.1960
0.5100, 0.0000, 0.3060, 0.1030, 0.1980, 0.1066, 0.5700, 0.3000, 0.5148, 0.1000, 0.2020
0.5700, 0.0000, 0.3010, 0.1170, 0.2020, 0.1396, 0.4200, 0.5000, 0.4625, 0.1200, 0.1550
0.5900, 1.0000, 0.2470, 0.1140, 0.1520, 0.1048, 0.2900, 0.5000, 0.4511, 0.0880, 0.0770
0.5100, 0.0000, 0.2770, 0.0990, 0.2290, 0.1456, 0.6900, 0.3000, 0.4277, 0.0770, 0.1910
0.7400, 0.0000, 0.2980, 0.1010, 0.1710, 0.1048, 0.5000, 0.3000, 0.4394, 0.0860, 0.0700
0.6700, 0.0000, 0.2670, 0.1050, 0.2250, 0.1354, 0.6900, 0.3000, 0.4635, 0.0960, 0.0730
0.4900, 0.0000, 0.1980, 0.0880, 0.1880, 0.1148, 0.5700, 0.3000, 0.4394, 0.0930, 0.0490
0.5700, 0.0000, 0.2330, 0.0880, 0.1550, 0.0636, 0.7800, 0.2000, 0.4205, 0.0780, 0.0650
0.5600, 1.0000, 0.3510, 0.1230, 0.1640, 0.0950, 0.3800, 0.4000, 0.5043, 0.1170, 0.2630
0.5200, 1.0000, 0.2970, 0.1090, 0.2280, 0.1628, 0.3100, 0.8000, 0.5142, 0.1030, 0.2480
0.6900, 0.0000, 0.2930, 0.1240, 0.2230, 0.1390, 0.5400, 0.4000, 0.5011, 0.1020, 0.2960
0.3700, 0.0000, 0.2030, 0.0830, 0.1850, 0.1246, 0.3800, 0.5000, 0.4719, 0.0880, 0.2140
0.2400, 0.0000, 0.2250, 0.0890, 0.1410, 0.0680, 0.5200, 0.3000, 0.4654, 0.0840, 0.1850
0.5500, 1.0000, 0.2270, 0.0930, 0.1540, 0.0942, 0.5300, 0.3000, 0.3526, 0.0750, 0.0780
0.3600, 0.0000, 0.2280, 0.0870, 0.1780, 0.1160, 0.4100, 0.4000, 0.4654, 0.0820, 0.0930
0.4200, 1.0000, 0.2400, 0.1070, 0.1500, 0.0850, 0.4400, 0.3000, 0.4654, 0.0960, 0.2520
0.2100, 0.0000, 0.2420, 0.0760, 0.1470, 0.0770, 0.5300, 0.3000, 0.4443, 0.0790, 0.1500
0.4100, 0.0000, 0.2020, 0.0620, 0.1530, 0.0890, 0.5000, 0.3000, 0.4249, 0.0890, 0.0770
0.5700, 1.0000, 0.2940, 0.1090, 0.1600, 0.0876, 0.3100, 0.5000, 0.5333, 0.0920, 0.2080
0.2000, 1.0000, 0.2210, 0.0870, 0.1710, 0.0996, 0.5800, 0.3000, 0.4205, 0.0780, 0.0770
0.6700, 1.0000, 0.2360, 0.1113, 0.1890, 0.1054, 0.7000, 0.2700, 0.4220, 0.0930, 0.1080
0.3400, 0.0000, 0.2520, 0.0770, 0.1890, 0.1206, 0.5300, 0.4000, 0.4344, 0.0790, 0.1600
0.4100, 1.0000, 0.2490, 0.0860, 0.1920, 0.1150, 0.6100, 0.3000, 0.4382, 0.0940, 0.0530
0.3800, 1.0000, 0.3300, 0.0780, 0.3010, 0.2150, 0.5000, 0.6020, 0.5193, 0.1080, 0.2200
0.5100, 0.0000, 0.2350, 0.1010, 0.1950, 0.1210, 0.5100, 0.4000, 0.4745, 0.0940, 0.1540
0.5200, 1.0000, 0.2640, 0.0913, 0.2180, 0.1520, 0.3900, 0.5590, 0.4905, 0.0990, 0.2590
0.6700, 0.0000, 0.2980, 0.0800, 0.1720, 0.0934, 0.6300, 0.3000, 0.4357, 0.0820, 0.0900
0.6100, 0.0000, 0.3000, 0.1080, 0.1940, 0.1000, 0.5200, 0.3730, 0.5347, 0.1050, 0.2460
0.6700, 1.0000, 0.2500, 0.1117, 0.1460, 0.0934, 0.3300, 0.4420, 0.4585, 0.1030, 0.1240
0.5600, 0.0000, 0.2700, 0.1050, 0.2470, 0.1606, 0.5400, 0.5000, 0.5088, 0.0940, 0.0670
0.6400, 0.0000, 0.2000, 0.0747, 0.1890, 0.1148, 0.6200, 0.3050, 0.4111, 0.0910, 0.0720
0.5800, 1.0000, 0.2550, 0.1120, 0.1630, 0.1106, 0.2900, 0.6000, 0.4762, 0.0860, 0.2570
0.5500, 0.0000, 0.2820, 0.0910, 0.2500, 0.1402, 0.6700, 0.4000, 0.5366, 0.1030, 0.2620
0.6200, 1.0000, 0.3330, 0.1140, 0.1820, 0.1140, 0.3800, 0.5000, 0.5011, 0.0960, 0.2750
0.5700, 1.0000, 0.2560, 0.0960, 0.2000, 0.1330, 0.5200, 0.3850, 0.4318, 0.1050, 0.1770
0.2000, 1.0000, 0.2420, 0.0880, 0.1260, 0.0722, 0.4500, 0.3000, 0.3784, 0.0740, 0.0710
0.5300, 1.0000, 0.2210, 0.0980, 0.1650, 0.1052, 0.4700, 0.4000, 0.4159, 0.0810, 0.0470
0.3200, 1.0000, 0.3140, 0.0890, 0.1530, 0.0842, 0.5600, 0.3000, 0.4159, 0.0900, 0.1870
0.4100, 0.0000, 0.2310, 0.0860, 0.1480, 0.0780, 0.5800, 0.3000, 0.4094, 0.0600, 0.1250
0.6000, 0.0000, 0.2340, 0.0767, 0.2470, 0.1480, 0.6500, 0.3800, 0.5136, 0.0770, 0.0780
0.2600, 0.0000, 0.1880, 0.0830, 0.1910, 0.1036, 0.6900, 0.3000, 0.4522, 0.0690, 0.0510
0.3700, 0.0000, 0.3080, 0.1120, 0.2820, 0.1972, 0.4300, 0.7000, 0.5342, 0.1010, 0.2580
0.4500, 0.0000, 0.3200, 0.1100, 0.2240, 0.1342, 0.4500, 0.5000, 0.5412, 0.0930, 0.2150
0.6700, 0.0000, 0.3160, 0.1160, 0.1790, 0.0904, 0.4100, 0.4000, 0.5472, 0.1000, 0.3030
0.3400, 1.0000, 0.3550, 0.1200, 0.2330, 0.1466, 0.3400, 0.7000, 0.5568, 0.1010, 0.2430
0.5000, 0.0000, 0.3190, 0.0783, 0.2070, 0.1492, 0.3800, 0.5450, 0.4595, 0.0840, 0.0910
0.7100, 0.0000, 0.2950, 0.0970, 0.2270, 0.1516, 0.4500, 0.5000, 0.5024, 0.1080, 0.1500
0.5700, 1.0000, 0.3160, 0.1170, 0.2250, 0.1076, 0.4000, 0.6000, 0.5958, 0.1130, 0.3100
0.4900, 0.0000, 0.2030, 0.0930, 0.1840, 0.1030, 0.6100, 0.3000, 0.4605, 0.0930, 0.1530
0.3500, 0.0000, 0.4130, 0.0810, 0.1680, 0.1028, 0.3700, 0.5000, 0.4949, 0.0940, 0.3460
0.4100, 1.0000, 0.2120, 0.1020, 0.1840, 0.1004, 0.6400, 0.3000, 0.4585, 0.0790, 0.0630
0.7000, 1.0000, 0.2410, 0.0823, 0.1940, 0.1492, 0.3100, 0.6260, 0.4234, 0.1050, 0.0890
0.5200, 0.0000, 0.2300, 0.1070, 0.1790, 0.1237, 0.4250, 0.4210, 0.4159, 0.0930, 0.0500
0.6000, 0.0000, 0.2560, 0.0780, 0.1950, 0.0954, 0.9100, 0.2000, 0.3761, 0.0870, 0.0390
0.6200, 0.0000, 0.2250, 0.1250, 0.2150, 0.0990, 0.9800, 0.2000, 0.4500, 0.0950, 0.1030
0.4400, 1.0000, 0.3820, 0.1230, 0.2010, 0.1266, 0.4400, 0.5000, 0.5024, 0.0920, 0.3080
0.2800, 1.0000, 0.1920, 0.0810, 0.1550, 0.0946, 0.5100, 0.3000, 0.3850, 0.0870, 0.1160
0.5800, 1.0000, 0.2900, 0.0850, 0.1560, 0.1092, 0.3600, 0.4000, 0.3989, 0.0860, 0.1450
0.3900, 1.0000, 0.2400, 0.0897, 0.1900, 0.1136, 0.5200, 0.3650, 0.4804, 0.1010, 0.0740
0.3400, 1.0000, 0.2060, 0.0980, 0.1830, 0.0920, 0.8300, 0.2000, 0.3689, 0.0920, 0.0450
0.6500, 0.0000, 0.2630, 0.0700, 0.2440, 0.1662, 0.5100, 0.5000, 0.4898, 0.0980, 0.1150
0.6600, 1.0000, 0.3460, 0.1150, 0.2040, 0.1394, 0.3600, 0.6000, 0.4963, 0.1090, 0.2640
0.5100, 0.0000, 0.2340, 0.0870, 0.2200, 0.1088, 0.9300, 0.2000, 0.4511, 0.0820, 0.0870
0.5000, 1.0000, 0.2920, 0.1190, 0.1620, 0.0852, 0.5400, 0.3000, 0.4736, 0.0950, 0.2020
0.5900, 1.0000, 0.2720, 0.1070, 0.1580, 0.1020, 0.3900, 0.4000, 0.4443, 0.0930, 0.1270
0.5200, 0.0000, 0.2700, 0.0783, 0.1340, 0.0730, 0.4400, 0.3050, 0.4443, 0.0690, 0.1820
0.6900, 1.0000, 0.2450, 0.1080, 0.2430, 0.1364, 0.4000, 0.6000, 0.5808, 0.1000, 0.2410
0.5300, 0.0000, 0.2410, 0.1050, 0.1840, 0.1134, 0.4600, 0.4000, 0.4812, 0.0950, 0.0660
0.4700, 1.0000, 0.2530, 0.0980, 0.1730, 0.1056, 0.4400, 0.4000, 0.4762, 0.1080, 0.0940
0.5200, 0.0000, 0.2880, 0.1130, 0.2800, 0.1740, 0.6700, 0.4000, 0.5273, 0.0860, 0.2830
0.3900, 0.0000, 0.2090, 0.0950, 0.1500, 0.0656, 0.6800, 0.2000, 0.4407, 0.0950, 0.0640
0.6700, 1.0000, 0.2300, 0.0700, 0.1840, 0.1280, 0.3500, 0.5000, 0.4654, 0.0990, 0.1020
0.5900, 1.0000, 0.2410, 0.0960, 0.1700, 0.0986, 0.5400, 0.3000, 0.4466, 0.0850, 0.2000
0.5100, 1.0000, 0.2810, 0.1060, 0.2020, 0.1222, 0.5500, 0.4000, 0.4820, 0.0870, 0.2650
0.2300, 1.0000, 0.1800, 0.0780, 0.1710, 0.0960, 0.4800, 0.4000, 0.4905, 0.0920, 0.0940
0.6800, 0.0000, 0.2590, 0.0930, 0.2530, 0.1812, 0.5300, 0.5000, 0.4543, 0.0980, 0.2300
0.4400, 0.0000, 0.2150, 0.0850, 0.1570, 0.0922, 0.5500, 0.3000, 0.3892, 0.0840, 0.1810
0.6000, 1.0000, 0.2430, 0.1030, 0.1410, 0.0866, 0.3300, 0.4000, 0.4673, 0.0780, 0.1560
0.5200, 0.0000, 0.2450, 0.0900, 0.1980, 0.1290, 0.2900, 0.7000, 0.5298, 0.0860, 0.2330
0.3800, 0.0000, 0.2130, 0.0720, 0.1650, 0.0602, 0.8800, 0.2000, 0.4431, 0.0900, 0.0600
0.6100, 0.0000, 0.2580, 0.0900, 0.2800, 0.1954, 0.5500, 0.5000, 0.4997, 0.0900, 0.2190
0.6800, 1.0000, 0.2480, 0.1010, 0.2210, 0.1514, 0.6000, 0.4000, 0.3871, 0.0870, 0.0800
0.2800, 1.0000, 0.3150, 0.0830, 0.2280, 0.1494, 0.3800, 0.6000, 0.5313, 0.0830, 0.0680
0.6500, 1.0000, 0.3350, 0.1020, 0.1900, 0.1262, 0.3500, 0.5000, 0.4970, 0.1020, 0.3320
0.6900, 0.0000, 0.2810, 0.1130, 0.2340, 0.1428, 0.5200, 0.4000, 0.5278, 0.0770, 0.2480
0.5100, 0.0000, 0.2430, 0.0853, 0.1530, 0.0716, 0.7100, 0.2150, 0.3951, 0.0820, 0.0840
0.2900, 0.0000, 0.3500, 0.0983, 0.2040, 0.1426, 0.5000, 0.4080, 0.4043, 0.0910, 0.2000
0.5500, 1.0000, 0.2350, 0.0930, 0.1770, 0.1268, 0.4100, 0.4000, 0.3829, 0.0830, 0.0550
0.3400, 1.0000, 0.3000, 0.0830, 0.1850, 0.1072, 0.5300, 0.3000, 0.4820, 0.0920, 0.0850
0.6700, 0.0000, 0.2070, 0.0830, 0.1700, 0.0998, 0.5900, 0.3000, 0.4025, 0.0770, 0.0890
0.4900, 0.0000, 0.2560, 0.0760, 0.1610, 0.0998, 0.5100, 0.3000, 0.3932, 0.0780, 0.0310
0.5500, 1.0000, 0.2290, 0.0810, 0.1230, 0.0672, 0.4100, 0.3000, 0.4304, 0.0880, 0.1290
0.5900, 1.0000, 0.2510, 0.0900, 0.1630, 0.1014, 0.4600, 0.4000, 0.4357, 0.0910, 0.0830
0.5300, 0.0000, 0.3320, 0.0827, 0.1860, 0.1068, 0.4600, 0.4040, 0.5112, 0.1020, 0.2750
0.4800, 1.0000, 0.2410, 0.1100, 0.2090, 0.1346, 0.5800, 0.4000, 0.4407, 0.1000, 0.0650
0.5200, 0.0000, 0.2950, 0.1043, 0.2110, 0.1328, 0.4900, 0.4310, 0.4984, 0.0980, 0.1980
0.6900, 0.0000, 0.2960, 0.1220, 0.2310, 0.1284, 0.5600, 0.4000, 0.5451, 0.0860, 0.2360
0.6000, 1.0000, 0.2280, 0.1100, 0.2450, 0.1898, 0.3900, 0.6000, 0.4394, 0.0880, 0.2530
0.4600, 1.0000, 0.2270, 0.0830, 0.1830, 0.1258, 0.3200, 0.6000, 0.4836, 0.0750, 0.1240
0.5100, 1.0000, 0.2620, 0.1010, 0.1610, 0.0996, 0.4800, 0.3000, 0.4205, 0.0880, 0.0440
0.6700, 1.0000, 0.2350, 0.0960, 0.2070, 0.1382, 0.4200, 0.5000, 0.4898, 0.1110, 0.1720
0.4900, 0.0000, 0.2210, 0.0850, 0.1360, 0.0634, 0.6200, 0.2190, 0.3970, 0.0720, 0.1140
0.4600, 1.0000, 0.2650, 0.0940, 0.2470, 0.1602, 0.5900, 0.4000, 0.4935, 0.1110, 0.1420
0.4700, 0.0000, 0.3240, 0.1050, 0.1880, 0.1250, 0.4600, 0.4090, 0.4443, 0.0990, 0.1090
0.7500, 0.0000, 0.3010, 0.0780, 0.2220, 0.1542, 0.4400, 0.5050, 0.4779, 0.0970, 0.1800
0.2800, 0.0000, 0.2420, 0.0930, 0.1740, 0.1064, 0.5400, 0.3000, 0.4220, 0.0840, 0.1440
0.6500, 1.0000, 0.3130, 0.1100, 0.2130, 0.1280, 0.4700, 0.5000, 0.5247, 0.0910, 0.1630
0.4200, 0.0000, 0.3010, 0.0910, 0.1820, 0.1148, 0.4900, 0.4000, 0.4511, 0.0820, 0.1470
0.5100, 0.0000, 0.2450, 0.0790, 0.2120, 0.1286, 0.6500, 0.3000, 0.4522, 0.0910, 0.0970
0.5300, 1.0000, 0.2770, 0.0950, 0.1900, 0.1018, 0.4100, 0.5000, 0.5464, 0.1010, 0.2200
0.5400, 0.0000, 0.2320, 0.1107, 0.2380, 0.1628, 0.4800, 0.4960, 0.4913, 0.1080, 0.1900
0.7300, 0.0000, 0.2700, 0.1020, 0.2110, 0.1210, 0.6700, 0.3000, 0.4745, 0.0990, 0.1090
0.5400, 0.0000, 0.2680, 0.1080, 0.1760, 0.0806, 0.6700, 0.3000, 0.4956, 0.1060, 0.1910
0.4200, 0.0000, 0.2920, 0.0930, 0.2490, 0.1742, 0.4500, 0.6000, 0.5004, 0.0920, 0.1220
0.7500, 0.0000, 0.3120, 0.1177, 0.2290, 0.1388, 0.2900, 0.7900, 0.5724, 0.1060, 0.2300
0.5500, 1.0000, 0.3210, 0.1127, 0.2070, 0.0924, 0.2500, 0.8280, 0.6105, 0.1110, 0.2420
0.6800, 1.0000, 0.2570, 0.1090, 0.2330, 0.1126, 0.3500, 0.7000, 0.6057, 0.1050, 0.2480
0.5700, 0.0000, 0.2690, 0.0980, 0.2460, 0.1652, 0.3800, 0.7000, 0.5366, 0.0960, 0.2490
0.4800, 0.0000, 0.3140, 0.0753, 0.2420, 0.1516, 0.3800, 0.6370, 0.5568, 0.1030, 0.1920
0.6100, 1.0000, 0.2560, 0.0850, 0.1840, 0.1162, 0.3900, 0.5000, 0.4970, 0.0980, 0.1310
0.6900, 0.0000, 0.3700, 0.1030, 0.2070, 0.1314, 0.5500, 0.4000, 0.4635, 0.0900, 0.2370
0.3800, 0.0000, 0.3260, 0.0770, 0.1680, 0.1006, 0.4700, 0.4000, 0.4625, 0.0960, 0.0780
0.4500, 1.0000, 0.2120, 0.0940, 0.1690, 0.0968, 0.5500, 0.3000, 0.4454, 0.1020, 0.1350
0.5100, 1.0000, 0.2920, 0.1070, 0.1870, 0.1390, 0.3200, 0.6000, 0.4382, 0.0950, 0.2440
0.7100, 1.0000, 0.2400, 0.0840, 0.1380, 0.0858, 0.3900, 0.4000, 0.4190, 0.0900, 0.1990
0.5700, 0.0000, 0.3610, 0.1170, 0.1810, 0.1082, 0.3400, 0.5000, 0.5268, 0.1000, 0.2700
0.5600, 1.0000, 0.2580, 0.1030, 0.1770, 0.1144, 0.3400, 0.5000, 0.4963, 0.0990, 0.1640
0.3200, 1.0000, 0.2200, 0.0880, 0.1370, 0.0786, 0.4800, 0.3000, 0.3951, 0.0780, 0.0720
0.5000, 0.0000, 0.2190, 0.0910, 0.1900, 0.1112, 0.6700, 0.3000, 0.4078, 0.0770, 0.0960
0.4300, 0.0000, 0.3430, 0.0840, 0.2560, 0.1726, 0.3300, 0.8000, 0.5529, 0.1040, 0.3060
0.5400, 1.0000, 0.2520, 0.1150, 0.1810, 0.1200, 0.3900, 0.5000, 0.4701, 0.0920, 0.0910
0.3100, 0.0000, 0.2330, 0.0850, 0.1900, 0.1308, 0.4300, 0.4000, 0.4394, 0.0770, 0.2140
0.5600, 0.0000, 0.2570, 0.0800, 0.2440, 0.1516, 0.5900, 0.4000, 0.5118, 0.0950, 0.0950
0.4400, 0.0000, 0.2510, 0.1330, 0.1820, 0.1130, 0.5500, 0.3000, 0.4249, 0.0840, 0.2160
0.5700, 1.0000, 0.3190, 0.1110, 0.1730, 0.1162, 0.4100, 0.4000, 0.4369, 0.0870, 0.2630

Test data:


# diabetes_norm_test_100.txt
#
0.6400, 1.0000, 0.2840, 0.1110, 0.1840, 0.1270, 0.4100, 0.4000, 0.4382, 0.0970, 0.1780
0.4300, 0.0000, 0.2810, 0.1210, 0.1920, 0.1210, 0.6000, 0.3000, 0.4007, 0.0930, 0.1130
0.1900, 0.0000, 0.2530, 0.0830, 0.2250, 0.1566, 0.4600, 0.5000, 0.4719, 0.0840, 0.2000
0.7100, 1.0000, 0.2610, 0.0850, 0.2200, 0.1524, 0.4700, 0.5000, 0.4635, 0.0910, 0.1390
0.5000, 1.0000, 0.2800, 0.1040, 0.2820, 0.1968, 0.4400, 0.6000, 0.5328, 0.0950, 0.1390
0.5900, 1.0000, 0.2360, 0.0730, 0.1800, 0.1074, 0.5100, 0.4000, 0.4682, 0.0840, 0.0880
0.5700, 0.0000, 0.2450, 0.0930, 0.1860, 0.0966, 0.7100, 0.3000, 0.4522, 0.0910, 0.1480
0.4900, 1.0000, 0.2100, 0.0820, 0.1190, 0.0854, 0.2300, 0.5000, 0.3970, 0.0740, 0.0880
0.4100, 1.0000, 0.3200, 0.1260, 0.1980, 0.1042, 0.4900, 0.4000, 0.5412, 0.1240, 0.2430
0.2500, 1.0000, 0.2260, 0.0850, 0.1300, 0.0710, 0.4800, 0.3000, 0.4007, 0.0810, 0.0710
0.5200, 1.0000, 0.1970, 0.0810, 0.1520, 0.0534, 0.8200, 0.2000, 0.4419, 0.0820, 0.0770
0.3400, 0.0000, 0.2120, 0.0840, 0.2540, 0.1134, 0.5200, 0.5000, 0.6094, 0.0920, 0.1090
0.4200, 1.0000, 0.3060, 0.1010, 0.2690, 0.1722, 0.5000, 0.5000, 0.5455, 0.1060, 0.2720
0.2800, 1.0000, 0.2550, 0.0990, 0.1620, 0.1016, 0.4600, 0.4000, 0.4277, 0.0940, 0.0600
0.4700, 1.0000, 0.2330, 0.0900, 0.1950, 0.1258, 0.5400, 0.4000, 0.4331, 0.0730, 0.0540
0.3200, 1.0000, 0.3100, 0.1000, 0.1770, 0.0962, 0.4500, 0.4000, 0.5187, 0.0770, 0.2210
0.4300, 0.0000, 0.1850, 0.0870, 0.1630, 0.0936, 0.6100, 0.2670, 0.3738, 0.0800, 0.0900
0.5900, 1.0000, 0.2690, 0.1040, 0.1940, 0.1266, 0.4300, 0.5000, 0.4804, 0.1060, 0.3110
0.5300, 0.0000, 0.2830, 0.1010, 0.1790, 0.1070, 0.4800, 0.4000, 0.4788, 0.1010, 0.2810
0.6000, 0.0000, 0.2570, 0.1030, 0.1580, 0.0846, 0.6400, 0.2000, 0.3850, 0.0970, 0.1820
0.5400, 1.0000, 0.3610, 0.1150, 0.1630, 0.0984, 0.4300, 0.4000, 0.4682, 0.1010, 0.3210
0.3500, 1.0000, 0.2410, 0.0947, 0.1550, 0.0974, 0.3200, 0.4840, 0.4852, 0.0940, 0.0580
0.4900, 1.0000, 0.2580, 0.0890, 0.1820, 0.1186, 0.3900, 0.5000, 0.4804, 0.1150, 0.2620
0.5800, 0.0000, 0.2280, 0.0910, 0.1960, 0.1188, 0.4800, 0.4000, 0.4984, 0.1150, 0.2060
0.3600, 1.0000, 0.3910, 0.0900, 0.2190, 0.1358, 0.3800, 0.6000, 0.5421, 0.1030, 0.2330
0.4600, 1.0000, 0.4220, 0.0990, 0.2110, 0.1370, 0.4400, 0.5000, 0.5011, 0.0990, 0.2420
0.4400, 1.0000, 0.2660, 0.0990, 0.2050, 0.1090, 0.4300, 0.5000, 0.5580, 0.1110, 0.1230
0.4600, 0.0000, 0.2990, 0.0830, 0.1710, 0.1130, 0.3800, 0.4500, 0.4585, 0.0980, 0.1670
0.5400, 0.0000, 0.2100, 0.0780, 0.1880, 0.1074, 0.7000, 0.3000, 0.3970, 0.0730, 0.0630
0.6300, 1.0000, 0.2550, 0.1090, 0.2260, 0.1032, 0.4600, 0.5000, 0.5951, 0.0870, 0.1970
0.4100, 1.0000, 0.2420, 0.0900, 0.1990, 0.1236, 0.5700, 0.4000, 0.4522, 0.0860, 0.0710
0.2800, 0.0000, 0.2540, 0.0930, 0.1410, 0.0790, 0.4900, 0.3000, 0.4174, 0.0910, 0.1680
0.1900, 0.0000, 0.2320, 0.0750, 0.1430, 0.0704, 0.5200, 0.3000, 0.4635, 0.0720, 0.1400
0.6100, 1.0000, 0.2610, 0.1260, 0.2150, 0.1298, 0.5700, 0.4000, 0.4949, 0.0960, 0.2170
0.4800, 0.0000, 0.3270, 0.0930, 0.2760, 0.1986, 0.4300, 0.6420, 0.5148, 0.0910, 0.1210
0.5400, 1.0000, 0.2730, 0.1000, 0.2000, 0.1440, 0.3300, 0.6000, 0.4745, 0.0760, 0.2350
0.5300, 1.0000, 0.2660, 0.0930, 0.1850, 0.1224, 0.3600, 0.5000, 0.4890, 0.0820, 0.2450
0.4800, 0.0000, 0.2280, 0.1010, 0.1100, 0.0416, 0.5600, 0.2000, 0.4127, 0.0970, 0.0400
0.5300, 0.0000, 0.2880, 0.1117, 0.1450, 0.0872, 0.4600, 0.3150, 0.4078, 0.0850, 0.0520
0.2900, 1.0000, 0.1810, 0.0730, 0.1580, 0.0990, 0.4100, 0.4000, 0.4500, 0.0780, 0.1040
0.6200, 0.0000, 0.3200, 0.0880, 0.1720, 0.0690, 0.3800, 0.4000, 0.5784, 0.1000, 0.1320
0.5000, 1.0000, 0.2370, 0.0920, 0.1660, 0.0970, 0.5200, 0.3000, 0.4443, 0.0930, 0.0880
0.5800, 1.0000, 0.2360, 0.0960, 0.2570, 0.1710, 0.5900, 0.4000, 0.4905, 0.0820, 0.0690
0.5500, 1.0000, 0.2460, 0.1090, 0.1430, 0.0764, 0.5100, 0.3000, 0.4357, 0.0880, 0.2190
0.5400, 0.0000, 0.2260, 0.0900, 0.1830, 0.1042, 0.6400, 0.3000, 0.4304, 0.0920, 0.0720
0.3600, 0.0000, 0.2780, 0.0730, 0.1530, 0.1044, 0.4200, 0.4000, 0.3497, 0.0730, 0.2010
0.6300, 1.0000, 0.2410, 0.1110, 0.1840, 0.1122, 0.4400, 0.4000, 0.4935, 0.0820, 0.1100
0.4700, 1.0000, 0.2650, 0.0700, 0.1810, 0.1048, 0.6300, 0.3000, 0.4190, 0.0700, 0.0510
0.5100, 1.0000, 0.3280, 0.1120, 0.2020, 0.1006, 0.3700, 0.5000, 0.5775, 0.1090, 0.2770
0.4200, 0.0000, 0.1990, 0.0760, 0.1460, 0.0832, 0.5500, 0.3000, 0.3664, 0.0790, 0.0630
0.3700, 1.0000, 0.2360, 0.0940, 0.2050, 0.1388, 0.5300, 0.4000, 0.4190, 0.1070, 0.1180
0.2800, 0.0000, 0.2210, 0.0820, 0.1680, 0.1006, 0.5400, 0.3000, 0.4205, 0.0860, 0.0690
0.5800, 0.0000, 0.2810, 0.1110, 0.1980, 0.0806, 0.3100, 0.6000, 0.6068, 0.0930, 0.2730
0.3200, 0.0000, 0.2650, 0.0860, 0.1840, 0.1016, 0.5300, 0.4000, 0.4990, 0.0780, 0.2580
0.2500, 1.0000, 0.2350, 0.0880, 0.1430, 0.0808, 0.5500, 0.3000, 0.3584, 0.0830, 0.0430
0.6300, 0.0000, 0.2600, 0.0857, 0.1550, 0.0782, 0.4600, 0.3370, 0.5037, 0.0970, 0.1980
0.5200, 0.0000, 0.2780, 0.0850, 0.2190, 0.1360, 0.4900, 0.4000, 0.5136, 0.0750, 0.2420
0.6500, 1.0000, 0.2850, 0.1090, 0.2010, 0.1230, 0.4600, 0.4000, 0.5075, 0.0960, 0.2320
0.4200, 0.0000, 0.3060, 0.1210, 0.1760, 0.0928, 0.6900, 0.3000, 0.4263, 0.0890, 0.1750
0.5300, 0.0000, 0.2220, 0.0780, 0.1640, 0.0810, 0.7000, 0.2000, 0.4174, 0.1010, 0.0930
0.7900, 1.0000, 0.2330, 0.0880, 0.1860, 0.1284, 0.3300, 0.6000, 0.4812, 0.1020, 0.1680
0.4300, 0.0000, 0.3540, 0.0930, 0.1850, 0.1002, 0.4400, 0.4000, 0.5318, 0.1010, 0.2750
0.4400, 0.0000, 0.3140, 0.1150, 0.1650, 0.0976, 0.5200, 0.3000, 0.4344, 0.0890, 0.2930
0.6200, 1.0000, 0.3780, 0.1190, 0.1130, 0.0510, 0.3100, 0.4000, 0.5043, 0.0840, 0.2810
0.3300, 0.0000, 0.1890, 0.0700, 0.1620, 0.0918, 0.5900, 0.3000, 0.4025, 0.0580, 0.0720
0.5600, 0.0000, 0.3500, 0.0793, 0.1950, 0.1408, 0.4200, 0.4640, 0.4111, 0.0960, 0.1400
0.6600, 0.0000, 0.2170, 0.1260, 0.2120, 0.1278, 0.4500, 0.4710, 0.5278, 0.1010, 0.1890
0.3400, 1.0000, 0.2530, 0.1110, 0.2300, 0.1620, 0.3900, 0.6000, 0.4977, 0.0900, 0.1810
0.4600, 1.0000, 0.2380, 0.0970, 0.2240, 0.1392, 0.4200, 0.5000, 0.5366, 0.0810, 0.2090
0.5000, 0.0000, 0.3180, 0.0820, 0.1360, 0.0692, 0.5500, 0.2000, 0.4078, 0.0850, 0.1360
0.6900, 0.0000, 0.3430, 0.1130, 0.2000, 0.1238, 0.5400, 0.4000, 0.4710, 0.1120, 0.2610
0.3400, 0.0000, 0.2630, 0.0870, 0.1970, 0.1200, 0.6300, 0.3000, 0.4249, 0.0960, 0.1130
0.7100, 1.0000, 0.2700, 0.0933, 0.2690, 0.1902, 0.4100, 0.6560, 0.5242, 0.0930, 0.1310
0.4700, 0.0000, 0.2720, 0.0800, 0.2080, 0.1456, 0.3800, 0.6000, 0.4804, 0.0920, 0.1740
0.4100, 0.0000, 0.3380, 0.1233, 0.1870, 0.1270, 0.4500, 0.4160, 0.4318, 0.1000, 0.2570
0.3400, 0.0000, 0.3300, 0.0730, 0.1780, 0.1146, 0.5100, 0.3490, 0.4127, 0.0920, 0.0550
0.5100, 0.0000, 0.2410, 0.0870, 0.2610, 0.1756, 0.6900, 0.4000, 0.4407, 0.0930, 0.0840
0.4300, 0.0000, 0.2130, 0.0790, 0.1410, 0.0788, 0.5300, 0.3000, 0.3829, 0.0900, 0.0420
0.5500, 0.0000, 0.2300, 0.0947, 0.1900, 0.1376, 0.3800, 0.5000, 0.4277, 0.1060, 0.1460
0.5900, 1.0000, 0.2790, 0.1010, 0.2180, 0.1442, 0.3800, 0.6000, 0.5187, 0.0950, 0.2120
0.2700, 1.0000, 0.3360, 0.1100, 0.2460, 0.1566, 0.5700, 0.4000, 0.5088, 0.0890, 0.2330
0.5100, 1.0000, 0.2270, 0.1030, 0.2170, 0.1624, 0.3000, 0.7000, 0.4812, 0.0800, 0.0910
0.4900, 1.0000, 0.2740, 0.0890, 0.1770, 0.1130, 0.3700, 0.5000, 0.4905, 0.0970, 0.1110
0.2700, 0.0000, 0.2260, 0.0710, 0.1160, 0.0434, 0.5600, 0.2000, 0.4419, 0.0790, 0.1520
0.5700, 1.0000, 0.2320, 0.1073, 0.2310, 0.1594, 0.4100, 0.5630, 0.5030, 0.1120, 0.1200
0.3900, 1.0000, 0.2690, 0.0930, 0.1360, 0.0754, 0.4800, 0.3000, 0.4143, 0.0990, 0.0670
0.6200, 1.0000, 0.3460, 0.1200, 0.2150, 0.1292, 0.4300, 0.5000, 0.5366, 0.1230, 0.3100
0.3700, 0.0000, 0.2330, 0.0880, 0.2230, 0.1420, 0.6500, 0.3400, 0.4357, 0.0820, 0.0940
0.4600, 0.0000, 0.2110, 0.0800, 0.2050, 0.1444, 0.4200, 0.5000, 0.4533, 0.0870, 0.1830
0.6800, 1.0000, 0.2350, 0.1010, 0.1620, 0.0854, 0.5900, 0.3000, 0.4477, 0.0910, 0.0660
0.5100, 0.0000, 0.3150, 0.0930, 0.2310, 0.1440, 0.4900, 0.4700, 0.5252, 0.1170, 0.1730
0.4100, 0.0000, 0.2080, 0.0860, 0.2230, 0.1282, 0.8300, 0.3000, 0.4078, 0.0890, 0.0720
0.5300, 0.0000, 0.2650, 0.0970, 0.1930, 0.1224, 0.5800, 0.3000, 0.4143, 0.0990, 0.0490
0.4500, 0.0000, 0.2420, 0.0830, 0.1770, 0.1184, 0.4500, 0.4000, 0.4220, 0.0820, 0.0640
0.3300, 0.0000, 0.1950, 0.0800, 0.1710, 0.0854, 0.7500, 0.2000, 0.3970, 0.0800, 0.0480
0.6000, 1.0000, 0.2820, 0.1120, 0.1850, 0.1138, 0.4200, 0.4000, 0.4984, 0.0930, 0.1780
0.4700, 1.0000, 0.2490, 0.0750, 0.2250, 0.1660, 0.4200, 0.5000, 0.4443, 0.1020, 0.1040
0.6000, 1.0000, 0.2490, 0.0997, 0.1620, 0.1066, 0.4300, 0.3770, 0.4127, 0.0950, 0.1320
0.3600, 0.0000, 0.3000, 0.0950, 0.2010, 0.1252, 0.4200, 0.4790, 0.5130, 0.0850, 0.2200
0.3600, 0.0000, 0.1960, 0.0710, 0.2500, 0.1332, 0.9700, 0.3000, 0.4595, 0.0920, 0.0570

Posted in Machine Learning, Scikit | Leave a comment

Comparing Support Vector Regression versus Post-Training-Trimmed Kernel Ridge Regression Using Scikit

Posted on June 9, 2026 by jamesdmccaffrey

Bottom line: I did a short experiment to compare support vector regression (SVR) with post-training-trimmed kernel ridge regression (KRR). I used the scikit library SVR and KernelRidge modules. Because of the nearly infinite number of hyperparameter values involved, it wasn’t possible to draw a strong conclusion, but my results strongly suggest the two techniques are essentially the same, or at least very similar, from a practical point of view.

Bear with me. The goal of a machine learning regression problem is to predict a single numeric value. For example, a bank might want to predict a maximum loan amount based on applicant age, sex, annual income, debt, and so on.

Common regression techniques include linear regression, quadratic regression, nearest neighbors regression, kernel ridge regression, Gaussian process regression, kernel support vector regression, neural network regression, random forest regression, and gradient boost regression. Each technique has dozens of variations, and each technique has pros and cons.

Two closely related machine learning regression techniques are kernel ridge regression (KRR) and support vector regression (SVR). Both techniques use a kernel function (usually the radial basis function, RBF) to compare two data items for similarity. Both techniques must store training data in order to make predictions. KRR must store all training items, while SVR eliminates some of the items during training, leaving just the “support vectors” that need to be stored. So, SVR uses less memory than KRR.

On the other hand, the loss function used by SVR is not differentiable, so SVR cannot be trained using stochastic gradient descent, and therefore, SVR does not scale well to very large datasets. So, KRR is easier to train than SVR.

Note: In addition to the kernel support vector regression technique discussed in this blog post, there is a linear support vector regression technique which is essentially useless in practice.

The idea of combining KRR and SVR goes like this: Briefly, I train a KRR model as normal using all data. Then I identify training data items that are predicted “too well” and remove them, leaving just pseudo support vectors. The somewhat unobvious SVR idea is that items that are predicted too well don’t help model accuracy very much, and can lead to an overfitted model. After removing some training items, I retrain a new KRR model using only the reduced training data. This gives a trimmed/sparse KRR model that approximates an SVR model.

In my mind, this technique gives the advantages of KRR (ability to handle very large datasets via SGD training) and the advantages of SVR (fewer stored model items and weights than KRR).

I first implemented the idea using the scikit KernelRidge module and the idea seemed to work well. But I wanted to take a second look, using a very small dataset, as a sanity check to make sure there weren’t any unexpected, bad, surprises. The output of one of the experiments is:

Begin SVR vs. trimmed KRR demo

Creating 20-item, 5-feature train data

X:
[[ 0.0458 -0.1872  1.4694 -1.4544  1.5328]
 [ 0.9008  0.4657  1.4883 -1.1651 -1.5362]
 [ 0.6536  0.8644  2.2698 -2.5530 -0.7422]
 [ 0.4002  0.9787  1.8676  1.7641  2.2409]
 [ 0.3869 -0.5108 -0.0282 -0.8955 -1.1806]
 [ 1.2303  1.2024 -0.3023  0.1563 -0.3873]
 [ 1.4543  0.7610  0.4439  0.1440  0.1217]
 [ 1.4941 -0.2052 -0.8541  0.3337  0.3131]
 [-0.3596 -0.8131  0.1774 -0.6725 -1.7263]
 [ 1.1788 -0.1799  1.0545  1.8959 -1.0708]
 [ 0.1290  1.1394  0.4023  0.7291 -1.2348]
 [ 0.9501 -0.1514  0.4106 -0.9773 -0.1032]
 [ 1.2224  0.2083  0.3564 -0.4032  0.9766]
 [-1.6302  0.4628  0.0519 -0.4018 -0.9073]
 [ 0.3782 -0.8878 -0.3479  0.1549 -1.9808]
 [ 0.0665  0.3025 -0.3627  0.4283 -0.6343]
 [ 0.0105  1.7859  0.4020  0.7066  0.1269]
 [-1.4200 -1.7063 -0.5097 -1.0486  1.9508]
 [-1.2528  0.7775 -0.2127 -0.4381 -1.6139]
 [-0.8708 -0.5788  0.0562 -0.6848 -0.3116]]

Creating and training SVR model
gamma = 0.1000
C = 1000.0000
epsilon = 8.0000
Done

Number support vectors: 10

Support vector indices:
[ 0  1  3  9 10 14 16 17 18 19]

Accuracy (within 0.10) train = 0.6000
MSE train = 64.0016
R2 train = 0.996314

===============================

Creating and training preliminary KRR RBF model
Setting gamma = 0.1000
Setting alpha = 0.0010
Done

Removing non-pseudo-support vectors
Using KRR trim epsilon = 0.1000

Number of pseudo support vectors = 10

Support vector indices:
[ 1  3  9 10 11 15 16 17 18 19]

Re-training trimmed KRR model
Done

Accuracy (within 0.10) train = 1.0000
MSE train = 0.0225
R2 train = 0.999999

End demo

The demo program begins by using the scikit make_regression() function to create a tiny synthetic dataset with just 20 rows/items, each with 5 predictor values. An SVR model ended up with 10 support vectors and an R2 score (essentially a scaled accuracy metric) of 0.9963 on the training data.

I tuned a trimmed KRR model to one with 10 pseudo-support vectors. The two techniques had 8 out of 10 support vectors in common — that’s good.

The trimmed KRR model had an R2 score of 0.9999 on the training data. The trimmed KRR model had better MSE than the SVR model, but that was expected because KRR minimizes MSE and SVR does not. The much better accuracy of the trimmed KRR model (100% for KRR vs. 60% for SVR) isn’t significant because the dataset is so small, and I didn’t use a test dataset to measure possible overfitting.

I noticed that both models were ultra-sensitive to hyperparameter values — gamma, C, and epsilon for SVR, and gamma, alpha, and trim-epsilon for trimmed KRR. This is a major weakness of both techniques.

Anyway, the sanity check was successful. It is possible to approximate a SVR regression model using a trimmed KRR model.

It’s common for me to refactor the software systems I create many times. It’s rarely possible to a get a non-trivial system completely correct on the first effort, and so a system is usually a collection of software sequels, so to speak, where each sequel is a bit better than its predecessor.

I’m a big fan of science fiction movies. There have been dozens of sci fi sequels. “Star Wars” (1977) and “Star Wars: The Empire Strikes Back” (1980). “Alien” (1979) and “Aliens” (1986). And so on. But most movie sequels are worse than their predecessor, however there are exceptions where the sequel is as-good-as, or even better than the original.

Left: In “The Quatermass Xperiment” aka “The Creeping Unknown” (1955), Dr. Bernard Quatermass leads a British effort to put men into space. Three astronauts are sent up, but only one man returns. He has been exposed to cosmic radiation and morphs into a deadly blob-like creature. The creature is eventually electrocuted. My personal grade = B+.

Right: In “Quatermass 2” aka “Enemy From Space” (1957), Quatermass discovers an alien invasion. The aliens use parasites to control the people in a small village and use the villagers to construct a plant to create alien food to support the invasion. Quatermass and the UK military blow up the plant and defeat the alien’s plans. My personal grade = A-.

Demo program. Replace the “lt” (less than) and “gt” with the Boolean operator symbols. (My blog editor chokes on symbols).

# svr_vs_krr_trimmed_scikit.py
# train using KRR, remove some, retrain using KRR

import numpy as np
from sklearn.kernel_ridge import KernelRidge
from sklearn.svm import SVR
from sklearn.datasets import make_regression

# KernelRidge(alpha=1, *, kernel='linear', gamma=None,
# degree=3, coef0=1, kernel_params=None)

# SVR(*, kernel='rbf', degree=3, gamma='scale',
# coef0=0.0, tol=0.001, C=1.0, epsilon=0.1,
# shrinking=True, cache_size=200, verbose=False,
# max_iter=-1)

# make_regression(n_samples=100, n_features=100,
# *, n_informative=10, n_targets=1, bias=0.0,
# effective_rank=None, tail_strength=0.5, noise=0.0,
# shuffle=True, coef=False, random_state=None)

# -----------------------------------------------------------

np.set_printoptions(precision=4, suppress=True,
  floatmode='fixed', linewidth=60)

# -----------------------------------------------------------

def accuracy(model, data_X, data_y, pct_close):
  n = len(data_X)
  n_correct = 0; n_wrong = 0
  for i in range(n):
    x = data_X[i].reshape(1,-1)
    y = data_y[i]
    y_pred = model.predict(x)[0]

    if np.abs(y - y_pred) "lt" np.abs(y * pct_close):
      n_correct += 1
    else: 
      n_wrong += 1
  return n_correct / (n_correct + n_wrong)

def mse(model, data_X, data_y):
  n = len(data_X)
  sum = 0.0
  for i in range(n):
    actual_y = data_y[i]
    pred_y = model.predict(data_X[i].reshape(1, -1))[0]
    diff = actual_y - pred_y
    sum += diff * diff
  return sum /n

# -----------------------------------------------------------
# -----------------------------------------------------------

print("\nBegin SVR vs. trimmmed KRR demo ")

np.set_printoptions(precision=4, suppress=True,
    floatmode='fixed')

print("\nCreating 20-item, 5-feature train data ")
X, y = make_regression(n_samples=20, n_features=5,
  n_informative=5, noise=0, random_state=0)

print("\nX: ")
print(X)

print("\nCreating and training SVR model ")
svr_gamma = 0.10
svr_C = 1000.0
svr_epsilon = 8.0
print("gamma = %0.4f " % svr_gamma)
print("C = %0.4f " % svr_C)
print("epsilon = %0.4f " % svr_epsilon)
svr_model = SVR(kernel='rbf', gamma=svr_gamma, C=svr_C,
  epsilon = svr_epsilon)
svr_model.fit(X, y)
print("Done ")

print("\nNumber support vectors: ", end="")
print(len(svr_model.support_))
print("\nSupport vector indices: ")
print(svr_model.support_)

X_sv = X[svr_model.support_,:]
y_sv = y[svr_model.support_]

acc_train = accuracy(svr_model, X_sv, y_sv, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
mse_train = mse(svr_model, X_sv, y_sv)
print("MSE train = %0.4f " % mse_train)
svr_r2 = svr_model.score(X_sv, y_sv)
print("R2 train = %0.6f " % svr_r2)

print("\n=============================== ")

print("\nCreating and training preliminary KRR model ")
krr_gamma = 0.1000
krr_alpha = 0.0010
print("Setting gamma = %0.4f " % krr_gamma)
print("Setting alpha = %0.4f " % krr_alpha)
model = KernelRidge(kernel='rbf', gamma=krr_gamma,
  alpha=krr_alpha)
model.fit(X, y)
print("Done ")

print("\nRemoving non-pseudo-support vectors ")
krr_trim_epsilon = 0.10
print("Using KRR trim epsilon = %0.4f " % krr_trim_epsilon)
# smaller epsilon = more support vectors
# larger epsilon = fewer support vectors

predictions = model.predict(X)
residuals = np.abs(y - predictions)
# keep only pts outside epsilon-tube (pseudo support vecs)
sv_indices = np.where(residuals "gt" krr_trim_epsilon)[0]
X_sv = X[sv_indices]
y_sv = y[sv_indices]
print("\nNumber of pseudo support vectors = " + \
  str(len(X_sv)))
print("\nSupport vector indices: ")
print(sv_indices)

# retrain
print("\nRe-training trimmed KRR model ")
krr_model = KernelRidge(kernel='rbf', 
  gamma=krr_gamma, alpha=krr_alpha)
krr_model.fit(X_sv, y_sv)
print("Done ")

acc_train = accuracy(krr_model, X_sv, y_sv, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
mse_train = mse(krr_model, X_sv, y_sv)
print("MSE train = %0.4f " % mse_train)
krr_r2 = krr_model.score(X_sv, y_sv)
print("R2 train = %0.6f " % krr_r2)

print("\nEnd demo ")

Posted in Machine Learning, Scikit | Leave a comment

Trimmed Kernel Ridge Regression to Approximate Support Vector Regression Using C#

Posted on June 8, 2026 by jamesdmccaffrey

As is often the case with interesting topics, explaining what the problem is, takes longer than explaining the solution. So bear with me.

The goal of a machine learning regression problem is to predict a single numeric value. For example, a bank might want to predict a maximum loan amount based on applicant age, sex, annual income, debt, and so on.

Two closely related machine learning regression techniques are kernel ridge regression (KRR) and support vector regression (SVR). Both techniques use a kernel function (usually the radial basis function, RBF) to compare two data items for similarity. Both techniques must store training data in order to make predictions, but KRR must store all training items, while SVR eliminates some of the items during training, leaving just the “support vectors” that need to be stored.

Note: In addition to the kernel support vector regression technique discussed in this blog post, there is a linear support vector regression technique which is essentially useless in practice.

KRR often gives slightly better prediction accuracy than SVR, is easier to train than SVR, can handle very large datasets via SGD training, and is much, much easier to implement than SVR.

One morning before work, I got the idea of combining KRR and SVR. Briefly, I train a KRR model as normal using all data. Then I identify training data items that are predicted “too well” and remove them, leaving just pseudo support vectors. The somewhat counterintuitive SVR idea is that items that are predicted too well don’t help model accuracy very much. Then I retrain a new KRR model using only the reduced training data. This gives a trimmed/sparse KRR model that approximates an SVR model.

In my mind, this technique gives the advantages of KRR (ability to handle very large datasets) and the advantages of SVR (fewer stored model items and weights than KRR) . . . if the technique works. I implemented the idea using the scikit KernelRidge module and the idea seemed to work well.

Let me point out that the idea is fairly obvious so I’m reasonably sure other people have used the idea, but I could find no examples or discussion on the Internet.

So, I decided to continue my investigation by refactoring the scikit Python code to the C# language. The effort was non-trivial.

The output of the C# demo is:


Begin C# kernel ridge regression approximation to SVR

Loading train (200) and test (40) data
Done

First three X predictors:
  -0.1660   0.4406  -0.9998  -0.3953  -0.7065
   0.0776  -0.1616   0.3704  -0.5911   0.7562
  -0.9452   0.3409  -0.1654   0.1174  -0.7192

First three target y:
  0.4840
  0.1568
  0.8054

Creating prelim, full KRR model
Setting RBF gamma = 0.3
Setting alpha decay =  0.00001
Done

Setting lrnRate = 0.0500
Setting maxEpochs = 5000
Setting auto-exit tol = 0.0010

Training model
epoch =      0 MSE = 0.0181 acc = 0.1700
Auto-exit at epoch 723
Done

Prelim model weights:
 -0.5698  -0.3515  -0.0638  . . . 
. . . 
 -0.0605   0.1449   0.6476

Evaluating full KRR model
Number model item/weights = 200

Train acc = 0.9800
Test acc = 0.9500

Train MSE = 0.0001
Test MSE = 0.0002

Begin trim/prune model
Using trim epsilon = 0.003500

Finding pseudo-support vectors
Done

Creating new, smaller train dataset
Done

Begin train new, smaller model
epoch =      0 MSE = 0.0215 acc = 0.1560
Auto-exit at epoch 768
Done

Evaluating new model
Number model items/weights = 109

Train acc = 0.9817
Test acc = 0.9500

Train MSE = 0.0001
Test MSE = 0.0002

End KRR trim/prune demo

The demo data is synthetic demo. There are 200 training items and 40 test items.

After training on 200 data items, those items that were predicted with a small error of less than 0.0035 were removed. There were 91 such well-predicted items, leaving a reduced training dataset of 109 items. After training a new model on the reduced dataset, the sparse model scored 98.17% accuracy on the reduced training data (107 out of 109 correct) and 95.00% accuracy on the test data (38 out of 40 correct).

To validate my C# trimmed KRR approximation to SVR, I ran the data through the scikit SVR module. A grid-search optimized scikit SVR model created 90 support vectors (somewhat better than the sparse KRR 109 items), with 96.50% accuracy on the training data (not quite as good as the C# trimmed KRR model), and 95.00% accuracy on the test data (identical to the sparse KRR model). In short, the C# trimmed KRR model and the SVR model are very similar. The large number of hyperparameters involved with KRR and SVR makes exact comparison impossible in practice.

When I get a chance, I’ll refactor the C# code so that the creation of the preliminary full KRR model, and the creation of a trimmed KRR model are all encapsulated into a single Train() function so the call will look like this:

KRRTrim model = new KRRTrim(gamma, alpha);
model.Train(trainX, trainY, lrnRate, exitTol, epsilon);

An interesting experiment.

Different regression techniques vary across several factors, including ability to handle complex data, ability to handle non-numeric data, predictive power, and explainability. Kernel ridge regression is moderately explainable — the model weights indicate which training items have the most influence.

I am a huge fan a 1950s science fiction movies. Two of the best science fiction films of all time feature scenes where a scientist uses a film projector to explain the threat to a roomful of officials and other scientists. These scenes were burned into my memory the first time I saw each film as a young man.

Left: In “Godzilla” (Japan 1954, U.S. 1956), Dr. Yamane has just returned to Tokyo from Odo island where he (and the audience) saw Godzilla for the first time. My personal grade = A+. Easily one of the top 10 sci fi movies of all time.

Right: In “Them!” (1954), Dr. Harold Medford and his daughter, Dr. Pat Medford, explain why giant ants have the potential to completely eradicate all humans within a matter of months. My personal grade = solid A.

The C# trimmed KRR demo program. Replace “lt” (less than), “gt”, “lte”, “gte” with Boolean operator symbols. (My blog editor chokes on symbols).

using System;
using System.IO;
using System.Collections.Generic;

namespace KernelRidgeRegressionSGDTrim
{
  internal class KernelRidgeRegressionSGDProgram
  {
    static void Main(string[] args)
    {
      Console.WriteLine("\nBegin C# kernel ridge " +
        "regression approximation to SVR ");

      Console.WriteLine("\nLoading train (200) and" +
        " test (40) data ");
      string trainFile =
        "..\\..\\..\\Data\\synthetic_train_200.txt";
      double[][] trainX =
        Utils.MatLoad(trainFile,
        new int[] { 0, 1, 2, 3, 4 }, ',', "#");
      double[] trainY =
        Utils.MatToVec(Utils.MatLoad(trainFile,
        new int[] { 5 }, ',', "#"));

      string testFile =
        "..\\..\\..\\Data\\synthetic_test_40.txt";
      double[][] testX =
        Utils.MatLoad(testFile,
        new int[] { 0, 1, 2, 3, 4 }, ',', "#");
      double[] testY =
        Utils.MatToVec(Utils.MatLoad(testFile,
        new int[] { 5 }, ',', "#"));
      Console.WriteLine("Done ");

      Console.WriteLine("\nFirst three X predictors: ");
      for (int i = 0; i "lt" 3; ++i)
        Utils.VecShow(trainX[i], 4, 9);
      Console.WriteLine("\nFirst three target y: ");
      for (int i = 0; i "lt" 3; ++i)
        Console.WriteLine(trainY[i].ToString("F4").
          PadLeft(8));

      Console.WriteLine("\nCreating prelim, " +
        "full KRR model ");
      double gamma = 0.3;    // RBF param
      double alpha = 1.0e-5;  // regularization
      Console.WriteLine("Setting RBF gamma = " +
        gamma.ToString("F1"));
      Console.WriteLine("Setting alpha decay =  " +
        alpha.ToString("F5"));
      KRR krr = new KRR(gamma, alpha);
      Console.WriteLine("Done ");

      double lrnRate = 0.05;
      int maxEpochs = 5000;
      double exitTol = 0.001;
      Console.WriteLine("\nSetting lrnRate = " +
        lrnRate.ToString("F4"));
      Console.WriteLine("Setting maxEpochs = " + maxEpochs);
      Console.WriteLine("Setting auto-exit tol = " +
        exitTol.ToString("F4"));
      Console.WriteLine("\nTraining model ");
      krr.Train(trainX, trainY, lrnRate, maxEpochs, exitTol);
      Console.WriteLine("Done ");
      Console.WriteLine("\nPrelim model weights: ");
      Utils.VecShow(krr.wts, 4, 9);

      Console.WriteLine("\nEvaluating full KRR model ");
      Console.WriteLine("Number model item/weights = " +
        krr.wts.Length);
      double trainAcc = krr.Accuracy(trainX, trainY, 0.10);
      double testAcc = krr.Accuracy(testX, testY, 0.10);

      Console.WriteLine("\nTrain acc = " +
        trainAcc.ToString("F4"));
      Console.WriteLine("Test acc = " +
        testAcc.ToString("F4"));

      double trainMSE = krr.MSE(trainX, trainY);
      double testMSE = krr.MSE(testX, testY);

      Console.WriteLine("\nTrain MSE = " +
        trainMSE.ToString("F4"));
      Console.WriteLine("Test MSE = " +
        testMSE.ToString("F4"));

      //double[] x = trainX[0];
      //Console.WriteLine("\nPredicting for trainX[0] ");
      //double y = krr.Predict(x);
      //Console.WriteLine("Predicted y = " +
      //  y.ToString("F4"));

      Console.WriteLine("\nBegin trim/prune model ");
      // larger epsilon = fewer retained weights
      // smaller epsilon = more retained weights
      double epsilon = 0.0035;
      Console.WriteLine("Using trim epsilon = " + 
        epsilon.ToString("F6"));
      Console.WriteLine("\nFinding pseudo-support vectors ");
      int[] isSupportVec = new int[trainX.Length];
      // 0 = not supp == predicts too well.
      // 1 = supp == predicts OK or weak.
      // find items and wts that predict 'too well'
      for (int i = 0; i "lt" trainX.Length; ++i)
      {
        if (Math.Abs(krr.Predict(trainX[i]) - trainY[i])
          "gt" epsilon)
          isSupportVec[i] = 1; // not a supper-accurate pred
      }
      Console.WriteLine("Done ");
      Console.WriteLine("\nCreating smaller train dataset ");
      // create a new, smaller training dataset
      List"lt"double[]"gt" tmpTrainX = 
        new List"lt"double[]"gt"();
      List"lt"double"gt" tmpTrainY = 
        new List"lt"double"gt"();
      for (int i = 0; i "lt" trainX.Length; ++i)
      {
        if (isSupportVec[i] == 1)
        {
          tmpTrainX.Add(trainX[i]);
          tmpTrainY.Add(trainY[i]);
        }
      }
      double[][] newTrainX = tmpTrainX.ToArray();
      double[] newTrainY = tmpTrainY.ToArray();
      Console.WriteLine("Done ");
      // train using new dataset
      Console.WriteLine("\nBegin train new, smaller model ");
      KRR newModel = new KRR(gamma, alpha);
      newModel.Train(newTrainX, newTrainY, lrnRate,
        maxEpochs, exitTol);
      Console.WriteLine("Done ");
      // analyze
      Console.WriteLine("\nEvaluating new model ");
      Console.WriteLine("Number model items/weights = " +
        newModel.wts.Length);

      trainAcc = newModel.Accuracy(newTrainX, newTrainY,
        0.10);
      testAcc = newModel.Accuracy(testX, testY, 0.10);

      Console.WriteLine("\nTrain acc = " +
        trainAcc.ToString("F4"));
      Console.WriteLine("Test acc = " +
        testAcc.ToString("F4"));

      trainMSE = newModel.MSE(newTrainX, newTrainY);
      testMSE = newModel.MSE(testX, testY);

      Console.WriteLine("\nTrain MSE = " +
        trainMSE.ToString("F4"));
      Console.WriteLine("Test MSE = " +
        testMSE.ToString("F4"));

      Console.WriteLine("\nEnd KRR trim/prune demo ");
      Console.ReadLine();
    } // Main()

  } // class Program

  /*
  scikit SVR version results for validation
  Creating scikit SVR model
  Setting gamma = 0.1000
  Setting C = 10.0000
  Setting epsilon = 0.0100

  Number support vectors = 90
  Accuracy (within 0.10) train = 0.9650
  Accuracy (within 0.10) test = 0.9500
  MSE train = 0.0001
  MSE test = 0.0002 
  */

  // ========================================================

  public class KRR
  {
    public double gamma;  // for RBF kernel
    public double alpha;  // regularization
    public double[][] trainX;  // need for prediction
    public double[] trainY;  // not needed this version
    public double[] wts;  // one per trainX item
    public Random rnd;

    // ------------------------------------------------------

    public KRR(double gamma, double alpha, int seed = 1)
    {
      this.gamma = gamma;
      this.alpha = alpha;
      this.rnd = new Random(seed);  // shuffle train order
    } // ctor

    // ------------------------------------------------------

    private void Shuffle(int[] indices)
    {
      // Fisher-Yates
      for (int i = 0; i "lt" indices.Length; ++i)
      {
        int ri = this.rnd.Next(i, indices.Length);
        int tmp = indices[i];
        indices[i] = indices[ri];
        indices[ri] = tmp;
      }
    } // Shuffle

    // ------------------------------------------------------

    public int Train(double[][] trainX, double[] trainY,
      double lrnRate, int maxEpochs, double exitTol)
    {
      // return epoch at exit
      int freq = maxEpochs / 5;  // when to show progress

      // 0. store trainX -- needed by Predict()
      this.trainX = trainX;  // by ref -- could copy
      this.trainY = trainY;  // not used this version

      this.wts = new double[trainX.Length];
      double lo = -0.10; double hi = 0.10;
      for (int i = 0; i "lt" this.wts.Length; ++i)
        this.wts[i] = (hi - lo) *
          this.rnd.NextDouble() + lo;

      // 1. set up indices for shuffling
      int[] indices = new int[trainX.Length];
      for (int i = 0; i "lt" indices.Length; ++i)
        indices[i] = i;

      // set up weights for auto-exit
      double[] prevWeights = new double[trainX.Length];
      for (int j = 0; j "lt" trainX.Length; ++j)
        prevWeights[j] = this.wts[j];

      for (int epoch = 0; epoch "lt" maxEpochs; ++epoch)
      {
        Shuffle(indices);
        for (int i = 0; i "lt" trainX.Length; ++i)
        {
          int idx = indices[i];
          double[] x = trainX[idx];
          double predY = this.Predict(x);
          double actualY = trainY[idx];

          // update wt assoc with x
          this.wts[idx] -= lrnRate * (predY - actualY);
        } // each item

        if (epoch % freq == 0)
        {
          double rmse = this.MSE(trainX, trainY);
          double acc = this.Accuracy(trainX, trainY, 0.10);
          string s1 = "epoch = " +
            epoch.ToString().PadLeft(6);
          string s2 = " MSE = " +
            rmse.ToString("F4");
          string s3 = " acc = " + acc.ToString("F4");
          Console.WriteLine(s1 + s2 + s3);
        }

        
        // if max_change_in_wts / max_weights "lt" tol
        int numWts = this.wts.Length;
        double[] weightDeltas = new double[numWts];
        for (int j = 0; j "lt" numWts; ++j)
          weightDeltas[j] = 
            Math.Abs(prevWeights[j] - this.wts[j]);
        double maxChange = weightDeltas[0];
        for (int j = 0; j "lt" numWts; ++j)
          if (weightDeltas[j] "gt" maxChange)
            maxChange = weightDeltas[j];

        double maxWeight = Math.Abs(this.wts[0]);
        for (int j = 0; j "lt" numWts; ++j)
          if (Math.Abs(this.wts[j]) "gt" maxWeight)
            maxWeight = Math.Abs(this.wts[j]);
        if (maxWeight != 0.0 &&
          (maxChange / maxWeight) "lt" exitTol)
        {
          Console.WriteLine("Auto-exit at epoch " + epoch);
          return epoch;
        }

        // an auto-exit didn't happen
        for (int j = 0; j "lt" numWts; ++j)
          prevWeights[j] = this.wts[j];
      
      } // each epoch

      // apply one final wt decay regularization
      // as opposed to small decay after every epoch
      for (int j = 0; j "lt" this.wts.Length; ++j)
        this.wts[j] *= (1.0 - this.alpha);
      return maxEpochs;

    } // Train

    // ------------------------------------------------------

    private double Rbf(double[] v1, double[] v2)
    {
      // the gamma version aot len_scale version
      int dim = v1.Length;
      double sum = 0.0;
      for (int i = 0; i "lt" dim; ++i)
      {
        sum += (v1[i] - v2[i]) * (v1[i] - v2[i]);
      }
      return Math.Exp(-1 * this.gamma * sum);
    }

    // ------------------------------------------------------

    public double Predict(double[] x)
    {
      int N = this.trainX.Length;
      double sum = 0.0;
      for (int i = 0; i "lt" N; ++i)
      {
        double[] xx = this.trainX[i];
        double k = this.Rbf(x, xx);
        sum += this.wts[i] * k;
      }
      return sum;
    }

    // ------------------------------------------------------

    public double Accuracy(double[][] dataX,
      double[] dataY, double pctClose)
    {
      int numCorrect = 0; int numWrong = 0;
      int n = dataX.Length;
      for (int i = 0; i "lt" n; ++i)
      {
        double[] x = dataX[i];
        double actualY = dataY[i];
        double predY = this.Predict(x);
        if (Math.Abs(actualY - predY) "lt"
          Math.Abs(actualY * pctClose))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / n;
    }

    // ------------------------------------------------------

    public double MSE(double[][] dataX, double[] dataY)
    {
      double sum = 0.0;
      int n = dataX.Length;
      for (int i = 0; i "lt" n; ++i)
      {
        double[] x = dataX[i];
        double actualY = dataY[i];
        double predY = this.Predict(x);
        sum += (actualY - predY) * (actualY - predY);
      }
      return sum / n;
    }
  } // class KRR

  // ========================================================

  public class Utils
  {
    // ------------------------------------------------------

    public static double[][] MatLoad(string fn,
      int[] usecols, char sep, string comment)
    {
      List"lt"double[]"gt" result = 
        new List"lt"double[]"gt"();
      string line = "";
      FileStream ifs = new FileStream(fn, FileMode.Open);
      StreamReader sr = new StreamReader(ifs);
      while ((line = sr.ReadLine()) != null)
      {
        if (line.StartsWith(comment) == true)
          continue;
        string[] tokens = line.Split(sep);
        List"lt"double"gt" lst = new List"lt"double"gt"();
        for (int j = 0; j "lt" usecols.Length; ++j)
          lst.Add(double.Parse(tokens[usecols[j]]));
        double[] row = lst.ToArray();
        result.Add(row);
      }
      sr.Close(); ifs.Close();
      return result.ToArray();
    }

    // ------------------------------------------------------

    public static double[] MatToVec(double[][] mat)
    {
      int nRows = mat.Length;
      int nCols = mat[0].Length;
      double[] result = new double[nRows * nCols];
      int k = 0;
      for (int i = 0; i "lt" nRows; ++i)
        for (int j = 0; j "lt" nCols; ++j)
          result[k++] = mat[i][j];
      return result;
    }

    // ------------------------------------------------------

    public static void MatShow(double[][] m, int dec, int wid)
    {
      int nRows = m.Length; int nCols = m[0].Length;
      double small = 1.0 / Math.Pow(10, dec);
      for (int i = 0; i "lt" nRows; ++i)
      {
        for (int j = 0; j "lt" nCols; ++j)
        {
          double v = m[i][j];
          if (Math.Abs(v) "lt" small) v = 0.0;
          Console.Write(v.ToString("F" + dec).
            PadLeft(wid));
        }
        Console.WriteLine("");
      }
    }

    // ------------------------------------------------------

    public static void VecShow(double[] vec, int dec, int wid)
    {
      for (int i = 0; i "lt" vec.Length; ++i)
        Console.Write(vec[i].ToString("F" + dec).
          PadLeft(wid));
      Console.WriteLine("");
    }

    // ------------------------------------------------------

  } // class Utils

  // ========================================================

} // ns

Training data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning | Leave a comment

Linear Regression With SGD Training and Early Exit From Scratch Using C#

Posted on June 5, 2026 by jamesdmccaffrey

I like to write code every day — it’s good mental exercise and just plain fun for me. One evening, after I gave my dogs their early evening walk, I figured I’d implement linear regression, from scratch, using C#, with SGD training, with an early exit condition out of the training loop.

There are two main forms of early exit from SGD training. The most common uses a validation dataset, and checks if the model weights aren’t changing significantly. This first form is usually called “early stopping”. I’ve always been skeptical of this technique (see https://jamesmccaffreyblog.com/2022/11/10/why-i-dont-use-neural-network-early-stopping/). The second form of early exit just checks the training data directly for no significant change in model weights. That’s the approach I used.

Briefly, in high-level pseudo-code:

loop max_epochs times
  shuffle order of training data
  loop each training item
    compute predicted y
    compute error = pred_y - target_y
    loop each model weight
      use error and learn_rate to update each weight
    end-loop
    update model bias
  end-loop
  after each epoch, 
    compute max_change_in_wts, max_wt
  if (max_change / max_wt less-than stop_tolerance)
    exit training loop
  end-if
end-loop
(at this point training done but no early exit happened)

The output of the demo is:

Begin C# linear regression SGD training with early exit

Loading synthetic train (200) and test (40) data
Done

First three train X:
 -0.1660  0.4406 -0.9998 -0.3953 -0.7065
  0.0776 -0.1616  0.3704 -0.5911  0.7562
 -0.9452  0.3409 -0.1654  0.1174 -0.7192

First three train y:
  0.4840
  0.1568
  0.8054

Setting lrnRate = 0.0005
Setting maxEpohcs = 500

Setting stop tol = 0.0001

Creating and training Linear Regression model
epoch =     0  MSE =   0.1331
epoch =   100  MSE =   0.0026
epoch =   200  MSE =   0.0026
Early exit at epoch 214
Done

Weights/coefficients:
-0.2654 0.0333 -0.0456 0.0357 -0.1145
Bias/constant: 0.3620

Evaluating model

Accuracy train (within 0.10) = 0.4600
Accuracy test (within 0.10) = 0.6250

MSE train = 0.0026
MSE test = 0.0020

Predicting for x =
  -0.1660   0.4406  -0.9998  -0.3953  -0.7065

Predicted y = 0.5330

End demo

There are many ways to decide when to early exit, but the one above is simple and usually effective. Good fun. Now it’s time to give my dogs their late night walk.

I’m a big fan of early science fiction art from the 1950s.

Left: By Alexander Leydenfrost (1888-1961).

Right: By Alex Schomburg (1905-1998).

Demo program. Replace “lt” (less than), “gt”, “lte”, “gte” with Boolean operator symbols (my blog editor chokes on symbols)

using System;
using System.IO;
using System.Collections.Generic;

namespace LinearRegressionSGDEarlyStop
{
  internal class LinearRegressionProgram
  {
    static void Main(string[] args)
    {
      Console.WriteLine("\nBegin C# linear regression" +
        " SGD training with early exit ");

      // 1. load data
      Console.WriteLine("\nLoading synthetic train" +
        " (200) and test (40) data");
      string trainFile =
        "..\\..\\..\\Data\\synthetic_train_200.txt";
      int[] colsX = new int[] { 0, 1, 2, 3, 4 };
      double[][] trainX =
        MatLoad(trainFile, colsX, ',', "#");
      double[] trainY =
        MatToVec(MatLoad(trainFile,
        new int[] { 5 }, ',', "#"));

      string testFile =
        "..\\..\\..\\Data\\synthetic_test_40.txt";
      double[][] testX =
        MatLoad(testFile, colsX, ',', "#");
      double[] testY =
        MatToVec(MatLoad(testFile,
        new int[] { 5 }, ',', "#"));
      Console.WriteLine("Done ");

      Console.WriteLine("\nFirst three train X: ");
      for (int i = 0; i "lt" 3; ++i)
        VecShow(trainX[i], 4, 8);

      Console.WriteLine("\nFirst three train y: ");
      for (int i = 0; i "lt" 3; ++i)
        Console.WriteLine(trainY[i].ToString("F4").
          PadLeft(8));

      // 2. create and train model
      double lrnRate = 0.0005;
      int maxEpochs = 500;
      double stopTol = 0.0001;
      //int seed = 0;
      Console.WriteLine("\nSetting lrnRate = " +
        lrnRate.ToString("F4"));
      Console.WriteLine("Setting maxEpohcs = " +
        maxEpochs);
      Console.WriteLine("\nSetting stop tol = " +
        stopTol.ToString("F4"));

      Console.WriteLine("\nCreating and training" +
        " Linear Regression model ");
      LinearRegressor model =
        new LinearRegressor();
      model.TrainSGD(trainX, trainY, lrnRate,
        maxEpochs, stopTol);
      Console.WriteLine("Done ");

      // 2b.show model parameters
      Console.WriteLine("\nWeights/coefficients: ");
      for (int i = 0; i "lt" model.weights.Length; ++i)
        Console.Write(model.weights[i].ToString("F4") + " ");
      Console.WriteLine("\nBias/constant: " +
        model.bias.ToString("F4"));

      // 3. evaluate model
      Console.WriteLine("\nEvaluating model ");

      double accTrain = model.Accuracy(trainX, trainY, 0.10);
      Console.WriteLine("\nAccuracy train (within 0.10) = " +
        accTrain.ToString("F4"));
      double accTest = model.Accuracy(testX, testY, 0.10);
      Console.WriteLine("Accuracy test (within 0.10) = " +
        accTest.ToString("F4"));

      double mseTrain = model.MSE(trainX, trainY);
      Console.WriteLine("\nMSE train = " +
        mseTrain.ToString("F4"));
      double mseTest = model.MSE(testX, testY);
      Console.WriteLine("MSE test = " +
        mseTest.ToString("F4"));

      // 4. use model
      double[] x = trainX[0];
      Console.WriteLine("\nPredicting for x = ");
      VecShow(x, 4, 9);
      double predY = model.Predict(x);
      Console.WriteLine("\nPredicted y = " +
        predY.ToString("F4"));

      Console.WriteLine("\nEnd demo ");
      Console.ReadLine();
    } // Main

    // ------------------------------------------------------
    // helpers for Main()
    // ------------------------------------------------------

    static double[][] MatLoad(string fn, int[] usecols,
      char sep, string comment)
    {
      List"lt"double[]"gt" result = 
        new List"lt"double[]"gt"();
      string line = "";
      FileStream ifs = new FileStream(fn, FileMode.Open);
      StreamReader sr = new StreamReader(ifs);
      while ((line = sr.ReadLine()) != null)
      {
        if (line.StartsWith(comment) == true)
          continue;
        string[] tokens = line.Split(sep);
        List"lt"double"gt" lst = new List"lt"double"gt"();
        for (int j = 0; j "lt" usecols.Length; ++j)
          lst.Add(double.Parse(tokens[usecols[j]]));
        double[] row = lst.ToArray();
        result.Add(row);
      }
      sr.Close(); ifs.Close();
      return result.ToArray();
    }

    static double[] MatToVec(double[][] M)
    {
      int nRows = M.Length;
      int nCols = M[0].Length;
      double[] result = new double[nRows * nCols];
      int k = 0;
      for (int i = 0; i "lt" nRows; ++i)
        for (int j = 0; j "lt" nCols; ++j)
          result[k++] = M[i][j];
      return result;
    }

    static void VecShow(double[] vec, int dec, int wid)
    {
      for (int i = 0; i "lt" vec.Length; ++i)
        Console.Write(vec[i].ToString("F" + dec).
          PadLeft(wid));
      Console.WriteLine("");
    }
  } // class Program

  public class LinearRegressor
  {
    public double[] weights;
    public double bias;
    private Random rnd;

    public LinearRegressor(int seed = 0)
    {
      this.weights = new double[0]; // keep compiler happy
      this.bias = 0;
      this.rnd = new Random(seed);
    }

    public int TrainSGD(double[][] trainX,
      double[] trainY, double lrnRate, int maxEpochs,
      double stopTol)
    {
      int n = trainX.Length;  int dim = trainX[0].Length;
      this.weights = new double[dim];

      // initialize weights and bias
      double low = -0.01; double hi = 0.01;
      for (int i = 0; i "lt" dim; ++i)
        this.weights[i] = (hi - low) *
          this.rnd.NextDouble() + low;
      this.bias = (hi - low) *
          this.rnd.NextDouble() + low;

      int[] indices = new int[n];  // of train data
      for (int i = 0; i "lt" n; ++i)
        indices[i] = i;

      double[] prevWeights = new double[dim];
      for (int j = 0; j "lt" dim; ++j)
        prevWeights[j] = this.weights[j];
      
      for (int epoch = 0; epoch "lt" maxEpochs; ++epoch)
      {
        Shuffle(indices, this.rnd);

        for (int i = 0; i "lt" n; ++i) // each train item
        {
          int ii = indices[i];
          double[] x = trainX[ii];
          double predY = this.Predict(x);
          double actualY = trainY[ii];
          for (int j = 0; j "lt" dim; ++j) // each weight
            this.weights[j] -= lrnRate *
              (predY - actualY) * x[j];
          this.bias -= lrnRate * (predY - actualY);
        }
        if (epoch % (int)(maxEpochs / 5) == 0) // progress
        {
          double mse = this.MSE(trainX, trainY);
          string s = "";
          s += "epoch = " + epoch.ToString().PadLeft(5);
          s += "  MSE = " + mse.ToString("F4").PadLeft(8);
          Console.WriteLine(s);
        }

        // check early-stop after each epoch
        // if max_change_in_wts / max_weights "lt" tol
        double[] weightDeltas = new double[dim];
        for (int j = 0; j "lt" dim; ++j)
          weightDeltas[j] = 
            Math.Abs(prevWeights[j] - this.weights[j]);
        double maxChange = weightDeltas[0];
        for (int j = 0; j "lt" dim; ++j)
          if (weightDeltas[j] "gt" maxChange)
            maxChange = weightDeltas[j];
        double maxWeight = this.weights[0];
        for (int j = 0; j "lt" dim; ++j)
          if (this.weights[j] "gt" maxWeight)
            maxWeight = this.weights[j];
        if (maxWeight != 0.0 &&
          (maxChange / maxWeight) "lt" stopTol)
        {
          Console.WriteLine("Early exit at epoch " + epoch);
          return epoch;
        }

        // early exit didn't happen
        for (int j = 0; j "lt" dim; ++j)
          prevWeights[j] = this.weights[j];

      } // epoch
      return maxEpochs;

    } // Train

    // ------------------------------------------------------

    public double Predict(double[] x)
    {
      double result = 0.0;
      for (int j = 0; j "lt" x.Length; ++j)
        result += x[j] * this.weights[j];
      result += this.bias;
      return result;
    }

    // ------------------------------------------------------

    public double Accuracy(double[][] dataX, double[] dataY,
      double pctClose)
    {
      int numCorrect = 0; int numWrong = 0;
      for (int i = 0; i "lt" dataX.Length; ++i)
      {
        double actualY = dataY[i];
        double predY = this.Predict(dataX[i]);
        if (Math.Abs(predY - actualY) "lt"
          (pctClose * actualY))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / (numWrong + numCorrect);
    }

    // ------------------------------------------------------

    public double MSE(double[][] dataX, double[] dataY)
    {
      int n = dataX.Length;
      double sum = 0.0;
      for (int i = 0; i "lt" n; ++i)
      {
        double actualY = dataY[i];
        double predY = this.Predict(dataX[i]);
        sum += (actualY - predY) * (actualY - predY);
      }
      return sum / n;
    }

    private static void Shuffle(int[] indices, Random rnd)
    {
      int n = indices.Length;
      for (int i = 0; i "lt" n; ++i)
      {
        int ri = rnd.Next(i, n);
        int tmp = indices[i];
        indices[i] = indices[ri];
        indices[ri] = tmp;
      }
    }

  } // class LinearRegressor

} // ns

Training data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning | Leave a comment

Why You Should Never Use Machine Learning Cross-Validation

Posted on June 4, 2026 by jamesdmccaffrey

I’ve worked with machine learning for decades. In my opinion, one of the least understood topics is cross-validation. I’ll cut to the chase and say that using cross-validation is never a good idea. Frankly, I am astonished that cross-validation techniques exists at all.

Well, let me temper all that by saying for the scenarios I work in, cross-validation is never useful. Some of my colleagues tell me they use cross-validation.

There are so many significant reasons not to use cross-validation, that listing them all would take several blog posts. But I’ll list just a few, any one of which is good reason to never use cross-validation.

To begin with, and set context, there are two categories of cross-validation: k-fold cross validation, and leave-one-out cross validation. Technically, leave-one-out is a specific type of k-fold, but in practice they are used differently. And then there are two possible reasons to use cross-validation: for evaluation of model generalizability, and for model hyperparameter tuning.

Let me dispense with leave-one-out cross validation. Suppose you have n training items, let’s say 200. For leave-one-out cross validation, you create and train 200 models using 199 training items and 1 test item.

Note: One exception to having to train n models is leave-one-out cross-validation for ridge regression. It’s possible to use a math trick to just train the model once instead of n times. But this trick only works for ridge regression, a relatively weak regression technique.

If you are trying to use the leave-one-out cross-validation for hyperparameter tuning, this makes no sense whatsoever, because all the models are being trained on almost exactly the same data (except for one item) and your evaluation metrics are all based on the MSE or R2 or whatever of a single data item.

If you are trying to use the leave-one-out cross-validation to measure model generalizability, you must average the MSE or R2 of 200 values, each of which was computed on a single data item, which makes no sense. And even if you did do this, you still need to measure generalizability using a hold-out test set that wasn’t part of the cross-validation process. I am stunned that leave-one-out cross validation exists.

OK, now, what about k-fold cross-validation. First problem: How do you choose k? Answer: You take a wild guess. I hope I don’t have to explain further.

For the sake of concreteness, assume you magically pick k=5 so that you have 5 subsets of the training data. Next question: Do you create one random partition set of 5 chunks of the data and then use those partitions (4 for training, 1 for testing/evaluation), or do you create different random partitions for each phase of the cross-validation. Answer: If you use one random partition, then if you have a bad selection, your parameter tuning results could be completely incorrect. In other words, you can have no confidence in the generated hyperparameters.

Using one fixed random partition for generalizability metrics is useless because you’ll still need to measure generalizability using a held-out test set. In other words, you can have no confidence in the generated generalizability metric.

Now if you use multiple random partitions, then there is a very strong likelihood (near certainty) that some data items will be used multiple times, and some data items won’t be used at all — so you will have no confidence in the generated generalizability metric.

I like the scikit-learn library a lot, but over time it has bloated out enormously. There are all kinds of modules that have cross-validation integrated — a bad idea. Here’s an example of ElasticNetCV — a particularly egregious example of bad bloat.

OK, so instead of using one random partition, or multiple random partitions, why not use stratified partitioning? Well, that’s another topic but briefly, stratified sampling has a whole host of problems that are just as bad, if not worse, than cross-validation problems. In short, if you use stratified partitions, you end up with hyperparameter or generalizability results with even less confidence.

At this point, while writing this post, I’m mentally exhausted because there are just so many reasons not to use any form of cross-validation, and there are several more, such as multiple problems with the inner-loop-for-hyper-parameter-tuning and outer-loop-for-generalizability-metric. And there are specific technical issues too, such as using cross-validation for learning rate tuning for iterative techniques, especially SGD. OK, I give up trying to explain why something that is so obviously bad is bad.

To summarize: 1.) It’s not even possible to set up cross-validation in a principled way, 2.) Leave-one-out cross-validation makes no sense for parameter tuning or generalizability metrics, 3.) Using k-fold validation produces hyperparameter or generalizability results that you can have absolutely no confidence in.

I get mildly annoyed that it’s even necessary to discuss why cross-validation shouldn’t be used. It’s like explaining why hitting yourself in the face with a brick isn’t a good idea.

I remember when I first was exposed to cross-validation, many years ago. I instantly knew it was bad. So, for year after year, I expected cross-validation to fade away from use. But no, cross-validation has lingered on and on, like lumpy mattresses in the rooms of a zero-star hotel.

Demo program. Replace the “lt” in accuracy() with the Boolean less-than operator.

# elastic_net_cv_scikit.py

import numpy as np
from sklearn.linear_model import ElasticNetCV

np.set_printoptions(precision=4, suppress=True,
  floatmode='fixed', linewidth=60)

# -----------------------------------------------------------

def accuracy(model, data_X, data_y, pct_close):
  n = len(data_X)
  n_correct = 0; n_wrong = 0
  for i in range(n):
    x = data_X[i].reshape(1,-1)
    y = data_y[i]
    y_pred = model.predict(x)[0]

    if np.abs(y - y_pred) "lt" np.abs(y * pct_close):
      n_correct += 1
    else: 
      n_wrong += 1
  return n_correct / (n_correct + n_wrong)

def mse(model, data_X, data_y):
  n = len(data_X)
  sum = 0.0
  for i in range(n):
    actual_y = data_y[i]
    pred_y = model.predict(data_X[i].reshape(1, -1))[0]
    diff = actual_y - pred_y
    sum += diff * diff
  return sum /n

# -----------------------------------------------------------

print("\nBegin elastic net CV regression using scikit ")
print("Note: Not a recommended technique ")

print("\nLoading train (200) and test (40) data ")
train_Xy = np.loadtxt(".\\Data\\synthetic_train_200.txt",
  usecols=[0,1,2,3,4,5], delimiter=",")
train_X = train_Xy[:,[0,1,2,3,4]]
train_y = train_Xy[:,5]

test_Xy = np.loadtxt(".\\Data\\synthetic_test_40.txt",
  usecols=[0,1,2,3,4,5], delimiter=",")
test_X = test_Xy[:,[0,1,2,3,4]]
test_y = test_Xy[:,5]

print("\nFirst three train X: ")
for i in range(3):
  print(train_X[i])
print("\nFirst three train y: ")
for i in range(3):
  print("%0.4f " % train_y[i])

# ElasticNetCV(*, l1_ratio=0.5, eps=0.001,
# n_alphas='deprecated', alphas='warn',
# fit_intercept=True, precompute='auto',
# max_iter=1000, tol=0.0001, cv=None, copy_X=True,
# verbose=0, n_jobs=None, positive=False,
# random_state=None, selection='cyclic')

print("\nCreating elastic-net CV with k=5 ")
n_folds = 5
model = ElasticNetCV(cv=n_folds, random_state=0)
print("Done ")

print("\nTraining model ")
model.fit(train_X, train_y)
print("Done. ")

print("\nModel weights: ")
print(model.coef_)
print("Model bias = %0.4f " % model.intercept_)

print("\nModel alphas examined: ")
print(model.alphas_)
print("Model selected alpha: %0.4f " % model.alpha_)

print("\nEvaluating model ")
acc_train = accuracy(model, train_X, train_y, 0.10)
acc_test = accuracy(model, test_X, test_y, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
print("Accuracy (within 0.10) test = %0.4f " % \
  acc_test)

mse_train = mse(model, train_X, train_y)
mse_test = mse(model, test_X, test_y)
print("\nMSE train = %0.6f " % mse_train)
print("MSE test = %0.6f " % mse_test)

x = train_X[0]
print("\nPredicting for x = ")
print(x)
pred_y = model.predict([x])[0]
print("Predicted y = %0.4f " % pred_y)

print("\nEnd demo ")

Training data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning, Scikit | Leave a comment

Support Vector Regression Using scikit SVR Applied to the Diabetes Dataset – Poor Results As Expected

Posted on June 3, 2026 by jamesdmccaffrey

I write code almost every day. Like many skills, writing code is something that must be practiced, and anyway, I enjoy writing code. One evening after work, I figured I’d run the well-known Diabetes Dataset through a support vector regression model. I used the scikit-learn SVR module.

Based on previous experiments with linear regression, quadratic regression, neural network regression, kernel ridge regression, random forest regression, and AdaBoost regression, I was almost certain that the scikit support vector regression model would give poor prediction accuracy — and that’s what happened.

The raw Diabetes Dataset looks like:

59, 2, 32.1, 101.00, 157,  93.2, 38, 4.00, 4.8598, 87, 151
48, 1, 21.6,  87.00, 183, 103.2, 70, 3.00, 3.8918, 69,  75
72, 2, 30.5,  93.00, 156,  93.6, 41, 4.00, 4.6728, 85, 141
. . .

The sex encoding isn’t explained anywhere but I suspect male = 1, female = 2 because there are 235 1 values and 206 2 values).

When using SVR, it’s a good idea to normalize predictor values so that a predictor with very large magnitude doesn’t overwhelm the other predictors.

0.5900, 1.0000, 0.3210, . . . 0.1510
0.4800, 0.0000, 0.2160, . . . 0.0750
0.7200, 1.0000, 0.3050, . . . 0.1410
. . .

I split the 442-items into a 342-item training set and a 100-item test set.

There are two versions of support vector regression. The kernel version is much more powerful than the linear version — so much so that linear SVR is essentially useless. I used kernel SVR.

Support vector regression is closely related to kernel ridge regression (KRR). Both techniques must store training items in memory to make a prediction. But KRR stores all training data, while SVR removes some of the items. On the other hand, SVR does not scale well to large datasets, but KRR can handle arbitrarily large datasets. Briefly, the SVR loss function is not differentiable so a SVR model must be trained with the very ugly quadratic programming technique. KRR can be trained using SGD which can handle any size data.

The output of the scikit SVR demo on the Diabetes Dataset is:

Begin scikit SVR on Diabetes Dataset demo

Loading diabetes train (342), test (100) data
Done

First three X predictors:
[0.5900 1.0000 0.3210 0.1010 0.1570 0.0932 0.3800 0.4000
 0.4860 0.0870]
[0.4800 0.0000 0.2160 0.0870 0.1830 0.1032 0.7000 0.3000
 0.3892 0.0690]
[0.7200 1.0000 0.3050 0.0930 0.1560 0.0936 0.4100 0.4000
 0.4673 0.0850]

First three y targets:
0.1510
0.0750
0.1410

Creating scikit SVR model
Setting gamma = 10.0000
Setting C = 1.0000
Setting epsilon = 0.0100
Done

Training SVR model
Done.

Number support vectors =
298

Evaluating model

Accuracy (within 0.10) train = 0.3538
Accuracy (within 0.10) test = 0.2283

MSE train = 0.0020
MSE test = 0.0034

End demo

These poor results were essentially the same as the results that I got using all other regression techniques.

I have done many experiments with the Diabetes Dataset and I’ve concluded the the default target value in the last column (a patient diabetes score) simply cannot be predicted well. But the variables in columns [4], [5], [6], [7], and [8] can be meaningfully predicted from the other columns.

In some sense, machine learning regression can be thought of as searching for hidden patterns in data.

Every cover of Playboy Magazine, except for the very first one (December 1953) has the company bunny logo somewhere. On most covers, the logo is clear and easy to see. But on some covers, the logo cleverly hidden. Looking for hidden items in a photo has an odd appeal to most people.

Left: On the cover of the issue from November 1990, the bunny logo is very difficult to spot. It’s incorporated into the model’s green and white pajama shirt near her right hand. I circled it in red so you don’t go crazy looking for it. Right: On the cover of the February 1991 issue, the bunny logo is disguised to look like one of the red hearts on the model’s shirt (I circled it in blue for you).

Demo program. Replace “lt” (less than) in the accuracy() function with the Boolean less-than operator symbol. (My blog editor chokes on symbols).

# diabetes_scikit_svr.py
# support vector regression for the Diabetes Dataset

import numpy as np
from sklearn.svm import SVR

# SVR(*, kernel='rbf', degree=3, gamma='scale',
# coef0=0.0, tol=0.001, C=1.0, epsilon=0.1, shrinking=True,
# cache_size=200, verbose=False, max_iter=-1)

# -----------------------------------------------------------

np.set_printoptions(precision=4, suppress=True,
  floatmode='fixed', linewidth=60)

# -----------------------------------------------------------

def accuracy(model, data_X, data_y, pct_close):
  n = len(data_X)
  n_correct = 0; n_wrong = 0
  for i in range(n):
    x = data_X[i].reshape(1,-1)
    y = data_y[i]
    y_pred = model.predict(x)[0]

    if np.abs(y - y_pred) "lt" np.abs(y * pct_close):
      n_correct += 1
    else: 
      n_wrong += 1
  return n_correct / (n_correct + n_wrong)

def mse(model, data_X, data_y):
  n = len(data_X)
  sum = 0.0
  for i in range(n):
    actual_y = data_y[i]
    pred_y = model.predict(data_X[i].reshape(1, -1))[0]
    diff = actual_y - pred_y
    sum += diff * diff
  return sum /n

# -----------------------------------------------------------

print("\nBegin scikit SVR on Diabetes Dataset demo ")

print("\nLoading diabetes train (342), test (100) data ")
train_file = ".\\Data\\diabetes_norm_train_342.txt"

cols_X = [0,1,2,3,4,5,6,7,8,9]  
col_y = 10  # cols # 4 5 6 7 8 are much better
train_X = np.loadtxt(train_file, comments="#",
  usecols=cols_X,
  delimiter=",",  dtype=np.float64)
train_y = np.loadtxt(train_file, comments="#", usecols=col_y,
  delimiter=",",  dtype=np.float64)

test_file = ".\\Data\\diabetes_norm_test_100.txt"
test_X = np.loadtxt(test_file, comments="#",
  usecols=cols_X,
  delimiter=",",  dtype=np.float64)
test_y = np.loadtxt(test_file, comments="#", usecols=col_y,
  delimiter=",",  dtype=np.float64)
print("Done ")

# alternative normalization and split
# from sklearn.datasets import load_diabetes
# from sklearn.model_selection import train_test_split
# X, y = load_diabetes(return_X_y=True, scaled=True)
# train_X, test_X, train_y, test_y = \
#   train_test_split(X, y, random_state=0)  # 25% test

print("\nFirst three X predictors: ")
for i in range(3):
  print(train_X[i])
print("\nFirst three y targets: ")
for i in range(3):
  print("%0.4f" % train_y[i])

# grid search
# gamma_vals = [0.01, 0.10, 1.0, 10.0]
# C_vals = [0.01, 0.10, 1.0, 10.0]
# epsilon_vals = [0.01, 0.10, 1.0, 10.0]
# for i in range(len(gamma_vals)):
#   for j in range(len(C_vals)):
#     for k in range(len(epsilon_vals)):
#       print("\n============")
#       print("gamma = %0.4f " % gamma_vals[i])
#       print("C = %0.4f " % C_vals[j])
#       print("epsilon = %0.4f " % epsilon_vals[k])
#       model = SVR(gamma=gamma_vals[i], C=C_vals[j],
#        epsilon=epsilon_vals[k])
#       model.fit(train_X, train_y)
#       acc_train = accuracy(model, train_X, train_y, 0.10)
#       print("Accuracy (within 0.10) train = %0.4f " % \
#        acc_train)
#       acc_test = accuracy(model, test_X, test_y, 0.10)
#       print("Accuracy (within 0.10) test = %0.4f " % \
#        acc_test)
#       mse_train = mse(model, train_X, train_y)
#       mse_test = mse(model, test_X, test_y)
#       print("MSE train = %0.4f " % mse_train)
#       print("MSE test = %0.4f " % mse_test)
#       print("\nNumber support vecs = " + \
#       str(len(model.support_)))
# best result:
# gamma = 10.0000
# C = 1.0000
# epsilon = 0.0100
# Accuracy (within 0.10) train = 0.3538
# Accuracy (within 0.10) test = 0.2283
# MSE train = 0.0020
# MSE test = 0.0034
# Number support vecs = 298

print("\nCreating scikit SVR model ")
gamma = 10.0000  # found by grid search (above)
C = 1.0000
epsilon = 0.0100
print("Setting gamma = %0.4f " % gamma)
print("Setting C = %0.4f " % C)
print("Setting epsilon = %0.4f " % epsilon)
model = SVR(kernel='rbf', gamma=gamma, C=C, epsilon=epsilon)
print("Done ")

print("\nTraining SVR model ")
model.fit(train_X, train_y)
print("Done. ")

# print("\nsupport vectors: ")
# print(model.support_)
print("\nNumber support vectors =  ")
print(len(model.support_))

print("\nEvaluating model ")
acc_train = accuracy(model, train_X, train_y, 0.10)
acc_test = accuracy(model, test_X, test_y, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
print("Accuracy (within 0.10) test = %0.4f " % \
  acc_test)

mse_train = mse(model, train_X, train_y)
mse_test = mse(model, test_X, test_y)
print("\nMSE train = %0.4f " % mse_train)
print("MSE test = %0.4f " % mse_test)

print("\nEnd demo ")

Training data:


# diabetes_norm_train_342.txt
# cols [0] to [9] predictors. col [10] target
# norm division constants:
# 100, -1, 100, 1000, 1000, 1000, 100, 10, 10, 1000, 1000
#
0.5900, 1.0000, 0.3210, 0.1010, 0.1570, 0.0932, 0.3800, 0.4000, 0.4860, 0.0870, 0.1510
0.4800, 0.0000, 0.2160, 0.0870, 0.1830, 0.1032, 0.7000, 0.3000, 0.3892, 0.0690, 0.0750
0.7200, 1.0000, 0.3050, 0.0930, 0.1560, 0.0936, 0.4100, 0.4000, 0.4673, 0.0850, 0.1410
0.2400, 0.0000, 0.2530, 0.0840, 0.1980, 0.1314, 0.4000, 0.5000, 0.4890, 0.0890, 0.2060
0.5000, 0.0000, 0.2300, 0.1010, 0.1920, 0.1254, 0.5200, 0.4000, 0.4291, 0.0800, 0.1350
0.2300, 0.0000, 0.2260, 0.0890, 0.1390, 0.0648, 0.6100, 0.2000, 0.4190, 0.0680, 0.0970
0.3600, 1.0000, 0.2200, 0.0900, 0.1600, 0.0996, 0.5000, 0.3000, 0.3951, 0.0820, 0.1380
0.6600, 1.0000, 0.2620, 0.1140, 0.2550, 0.1850, 0.5600, 0.4550, 0.4249, 0.0920, 0.0630
0.6000, 1.0000, 0.3210, 0.0830, 0.1790, 0.1194, 0.4200, 0.4000, 0.4477, 0.0940, 0.1100
0.2900, 0.0000, 0.3000, 0.0850, 0.1800, 0.0934, 0.4300, 0.4000, 0.5385, 0.0880, 0.3100
0.2200, 0.0000, 0.1860, 0.0970, 0.1140, 0.0576, 0.4600, 0.2000, 0.3951, 0.0830, 0.1010
0.5600, 1.0000, 0.2800, 0.0850, 0.1840, 0.1448, 0.3200, 0.6000, 0.3584, 0.0770, 0.0690
0.5300, 0.0000, 0.2370, 0.0920, 0.1860, 0.1092, 0.6200, 0.3000, 0.4304, 0.0810, 0.1790
0.5000, 1.0000, 0.2620, 0.0970, 0.1860, 0.1054, 0.4900, 0.4000, 0.5063, 0.0880, 0.1850
0.6100, 0.0000, 0.2400, 0.0910, 0.2020, 0.1154, 0.7200, 0.3000, 0.4291, 0.0730, 0.1180
0.3400, 1.0000, 0.2470, 0.1180, 0.2540, 0.1842, 0.3900, 0.7000, 0.5037, 0.0810, 0.1710
0.4700, 0.0000, 0.3030, 0.1090, 0.2070, 0.1002, 0.7000, 0.3000, 0.5215, 0.0980, 0.1660
0.6800, 1.0000, 0.2750, 0.1110, 0.2140, 0.1470, 0.3900, 0.5000, 0.4942, 0.0910, 0.1440
0.3800, 0.0000, 0.2540, 0.0840, 0.1620, 0.1030, 0.4200, 0.4000, 0.4443, 0.0870, 0.0970
0.4100, 0.0000, 0.2470, 0.0830, 0.1870, 0.1082, 0.6000, 0.3000, 0.4543, 0.0780, 0.1680
0.3500, 0.0000, 0.2110, 0.0820, 0.1560, 0.0878, 0.5000, 0.3000, 0.4511, 0.0950, 0.0680
0.2500, 1.0000, 0.2430, 0.0950, 0.1620, 0.0986, 0.5400, 0.3000, 0.3850, 0.0870, 0.0490
0.2500, 0.0000, 0.2600, 0.0920, 0.1870, 0.1204, 0.5600, 0.3000, 0.3970, 0.0880, 0.0680
0.6100, 1.0000, 0.3200, 0.1037, 0.2100, 0.0852, 0.3500, 0.6000, 0.6107, 0.1240, 0.2450
0.3100, 0.0000, 0.2970, 0.0880, 0.1670, 0.1034, 0.4800, 0.4000, 0.4357, 0.0780, 0.1840
0.3000, 1.0000, 0.2520, 0.0830, 0.1780, 0.1184, 0.3400, 0.5000, 0.4852, 0.0830, 0.2020
0.1900, 0.0000, 0.1920, 0.0870, 0.1240, 0.0540, 0.5700, 0.2000, 0.4174, 0.0900, 0.1370
0.4200, 0.0000, 0.3190, 0.0830, 0.1580, 0.0876, 0.5300, 0.3000, 0.4466, 0.1010, 0.0850
0.6300, 0.0000, 0.2440, 0.0730, 0.1600, 0.0914, 0.4800, 0.3000, 0.4635, 0.0780, 0.1310
0.6700, 1.0000, 0.2580, 0.1130, 0.1580, 0.0542, 0.6400, 0.2000, 0.5293, 0.1040, 0.2830
0.3200, 0.0000, 0.3050, 0.0890, 0.1820, 0.1106, 0.5600, 0.3000, 0.4344, 0.0890, 0.1290
0.4200, 0.0000, 0.2030, 0.0710, 0.1610, 0.0812, 0.6600, 0.2000, 0.4234, 0.0810, 0.0590
0.5800, 1.0000, 0.3800, 0.1030, 0.1500, 0.1072, 0.2200, 0.7000, 0.4644, 0.0980, 0.3410
0.5700, 0.0000, 0.2170, 0.0940, 0.1570, 0.0580, 0.8200, 0.2000, 0.4443, 0.0920, 0.0870
0.5300, 0.0000, 0.2050, 0.0780, 0.1470, 0.0842, 0.5200, 0.3000, 0.3989, 0.0750, 0.0650
0.6200, 1.0000, 0.2350, 0.0803, 0.2250, 0.1128, 0.8600, 0.2620, 0.4875, 0.0960, 0.1020
0.5200, 0.0000, 0.2850, 0.1100, 0.1950, 0.0972, 0.6000, 0.3000, 0.5242, 0.0850, 0.2650
0.4600, 0.0000, 0.2740, 0.0780, 0.1710, 0.0880, 0.5800, 0.3000, 0.4828, 0.0900, 0.2760
0.4800, 1.0000, 0.3300, 0.1230, 0.2530, 0.1636, 0.4400, 0.6000, 0.5425, 0.0970, 0.2520
0.4800, 1.0000, 0.2770, 0.0730, 0.1910, 0.1194, 0.4600, 0.4000, 0.4852, 0.0920, 0.0900
0.5000, 1.0000, 0.2560, 0.1010, 0.2290, 0.1622, 0.4300, 0.5000, 0.4779, 0.1140, 0.1000
0.2100, 0.0000, 0.2010, 0.0630, 0.1350, 0.0690, 0.5400, 0.3000, 0.4094, 0.0890, 0.0550
0.3200, 1.0000, 0.2540, 0.0903, 0.1530, 0.1004, 0.3400, 0.4500, 0.4533, 0.0830, 0.0610
0.5400, 0.0000, 0.2420, 0.0740, 0.2040, 0.1090, 0.8200, 0.2000, 0.4174, 0.1090, 0.0920
0.6100, 1.0000, 0.3270, 0.0970, 0.1770, 0.1184, 0.2900, 0.6000, 0.4997, 0.0870, 0.2590
0.5600, 1.0000, 0.2310, 0.1040, 0.1810, 0.1164, 0.4700, 0.4000, 0.4477, 0.0790, 0.0530
0.3300, 0.0000, 0.2530, 0.0850, 0.1550, 0.0850, 0.5100, 0.3000, 0.4554, 0.0700, 0.1900
0.2700, 0.0000, 0.1960, 0.0780, 0.1280, 0.0680, 0.4300, 0.3000, 0.4443, 0.0710, 0.1420
0.6700, 1.0000, 0.2250, 0.0980, 0.1910, 0.1192, 0.6100, 0.3000, 0.3989, 0.0860, 0.0750
0.3700, 1.0000, 0.2770, 0.0930, 0.1800, 0.1194, 0.3000, 0.6000, 0.5030, 0.0880, 0.1420
0.5800, 0.0000, 0.2570, 0.0990, 0.1570, 0.0916, 0.4900, 0.3000, 0.4407, 0.0930, 0.1550
0.6500, 1.0000, 0.2790, 0.1030, 0.1590, 0.0968, 0.4200, 0.4000, 0.4615, 0.0860, 0.2250
0.3400, 0.0000, 0.2550, 0.0930, 0.2180, 0.1440, 0.5700, 0.4000, 0.4443, 0.0880, 0.0590
0.4600, 0.0000, 0.2490, 0.1150, 0.1980, 0.1296, 0.5400, 0.4000, 0.4277, 0.1030, 0.1040
0.3500, 0.0000, 0.2870, 0.0970, 0.2040, 0.1268, 0.6400, 0.3000, 0.4190, 0.0930, 0.1820
0.3700, 0.0000, 0.2180, 0.0840, 0.1840, 0.1010, 0.7300, 0.3000, 0.3912, 0.0930, 0.1280
0.3700, 0.0000, 0.3020, 0.0870, 0.1660, 0.0960, 0.4000, 0.4150, 0.5011, 0.0870, 0.0520
0.4100, 0.0000, 0.2050, 0.0800, 0.1240, 0.0488, 0.6400, 0.2000, 0.4025, 0.0750, 0.0370
0.6000, 0.0000, 0.2040, 0.1050, 0.1980, 0.0784, 0.9900, 0.2000, 0.4635, 0.0790, 0.1700
0.6600, 1.0000, 0.2400, 0.0980, 0.2360, 0.1464, 0.5800, 0.4000, 0.5063, 0.0960, 0.1700
0.2900, 0.0000, 0.2600, 0.0830, 0.1410, 0.0652, 0.6400, 0.2000, 0.4078, 0.0830, 0.0610
0.3700, 1.0000, 0.2680, 0.0790, 0.1570, 0.0980, 0.2800, 0.6000, 0.5043, 0.0960, 0.1440
0.4100, 1.0000, 0.2570, 0.0830, 0.1810, 0.1066, 0.6600, 0.3000, 0.3738, 0.0850, 0.0520
0.3900, 0.0000, 0.2290, 0.0770, 0.2040, 0.1432, 0.4600, 0.4000, 0.4304, 0.0740, 0.1280
0.6700, 1.0000, 0.2400, 0.0830, 0.1430, 0.0772, 0.4900, 0.3000, 0.4431, 0.0940, 0.0710
0.3600, 1.0000, 0.2410, 0.1120, 0.1930, 0.1250, 0.3500, 0.6000, 0.5106, 0.0950, 0.1630
0.4600, 1.0000, 0.2470, 0.0850, 0.1740, 0.1232, 0.3000, 0.6000, 0.4644, 0.0960, 0.1500
0.6000, 1.0000, 0.2500, 0.0897, 0.1850, 0.1208, 0.4600, 0.4020, 0.4511, 0.0920, 0.0970
0.5900, 1.0000, 0.2360, 0.0830, 0.1650, 0.1000, 0.4700, 0.4000, 0.4500, 0.0920, 0.1600
0.5300, 0.0000, 0.2210, 0.0930, 0.1340, 0.0762, 0.4600, 0.3000, 0.4078, 0.0960, 0.1780
0.4800, 0.0000, 0.1990, 0.0910, 0.1890, 0.1096, 0.6900, 0.3000, 0.3951, 0.1010, 0.0480
0.4800, 0.0000, 0.2950, 0.1310, 0.2070, 0.1322, 0.4700, 0.4000, 0.4935, 0.1060, 0.2700
0.6600, 1.0000, 0.2600, 0.0910, 0.2640, 0.1466, 0.6500, 0.4000, 0.5568, 0.0870, 0.2020
0.5200, 1.0000, 0.2450, 0.0940, 0.2170, 0.1494, 0.4800, 0.5000, 0.4585, 0.0890, 0.1110
0.5200, 1.0000, 0.2660, 0.1110, 0.2090, 0.1264, 0.6100, 0.3000, 0.4682, 0.1090, 0.0850
0.4600, 1.0000, 0.2350, 0.0870, 0.1810, 0.1148, 0.4400, 0.4000, 0.4710, 0.0980, 0.0420
0.4000, 1.0000, 0.2900, 0.1150, 0.0970, 0.0472, 0.3500, 0.2770, 0.4304, 0.0950, 0.1700
0.2200, 0.0000, 0.2300, 0.0730, 0.1610, 0.0978, 0.5400, 0.3000, 0.3829, 0.0910, 0.2000
0.5000, 0.0000, 0.2100, 0.0880, 0.1400, 0.0718, 0.3500, 0.4000, 0.5112, 0.0710, 0.2520
0.2000, 0.0000, 0.2290, 0.0870, 0.1910, 0.1282, 0.5300, 0.4000, 0.3892, 0.0850, 0.1130
0.6800, 0.0000, 0.2750, 0.1070, 0.2410, 0.1496, 0.6400, 0.4000, 0.4920, 0.0900, 0.1430
0.5200, 1.0000, 0.2430, 0.0860, 0.1970, 0.1336, 0.4400, 0.5000, 0.4575, 0.0910, 0.0510
0.4400, 0.0000, 0.2310, 0.0870, 0.2130, 0.1264, 0.7700, 0.3000, 0.3871, 0.0720, 0.0520
0.3800, 0.0000, 0.2730, 0.0810, 0.1460, 0.0816, 0.4700, 0.3000, 0.4466, 0.0810, 0.2100
0.4900, 0.0000, 0.2270, 0.0653, 0.1680, 0.0962, 0.6200, 0.2710, 0.3892, 0.0600, 0.0650
0.6100, 0.0000, 0.3300, 0.0950, 0.1820, 0.1148, 0.5400, 0.3000, 0.4190, 0.0740, 0.1410
0.2900, 1.0000, 0.1940, 0.0830, 0.1520, 0.1058, 0.3900, 0.4000, 0.3584, 0.0830, 0.0550
0.6100, 0.0000, 0.2580, 0.0980, 0.2350, 0.1258, 0.7600, 0.3000, 0.5112, 0.0820, 0.1340
0.3400, 1.0000, 0.2260, 0.0750, 0.1660, 0.0918, 0.6000, 0.3000, 0.4263, 0.1080, 0.0420
0.3600, 0.0000, 0.2190, 0.0890, 0.1890, 0.1052, 0.6800, 0.3000, 0.4369, 0.0960, 0.1110
0.5200, 0.0000, 0.2400, 0.0830, 0.1670, 0.0866, 0.7100, 0.2000, 0.3850, 0.0940, 0.0980
0.6100, 0.0000, 0.3120, 0.0790, 0.2350, 0.1568, 0.4700, 0.5000, 0.5050, 0.0960, 0.1640
0.4300, 0.0000, 0.2680, 0.1230, 0.1930, 0.1022, 0.6700, 0.3000, 0.4779, 0.0940, 0.0480
0.3500, 0.0000, 0.2040, 0.0650, 0.1870, 0.1056, 0.6700, 0.2790, 0.4277, 0.0780, 0.0960
0.2700, 0.0000, 0.2480, 0.0910, 0.1890, 0.1068, 0.6900, 0.3000, 0.4190, 0.0690, 0.0900
0.2900, 0.0000, 0.2100, 0.0710, 0.1560, 0.0970, 0.3800, 0.4000, 0.4654, 0.0900, 0.1620
0.6400, 1.0000, 0.2730, 0.1090, 0.1860, 0.1076, 0.3800, 0.5000, 0.5308, 0.0990, 0.1500
0.4100, 0.0000, 0.3460, 0.0873, 0.2050, 0.1426, 0.4100, 0.5000, 0.4673, 0.1100, 0.2790
0.4900, 1.0000, 0.2590, 0.0910, 0.1780, 0.1066, 0.5200, 0.3000, 0.4575, 0.0750, 0.0920
0.4800, 0.0000, 0.2040, 0.0980, 0.2090, 0.1394, 0.4600, 0.5000, 0.4771, 0.0780, 0.0830
0.5300, 0.0000, 0.2800, 0.0880, 0.2330, 0.1438, 0.5800, 0.4000, 0.5050, 0.0910, 0.1280
0.5300, 1.0000, 0.2220, 0.1130, 0.1970, 0.1152, 0.6700, 0.3000, 0.4304, 0.1000, 0.1020
0.2300, 0.0000, 0.2900, 0.0900, 0.2160, 0.1314, 0.6500, 0.3000, 0.4585, 0.0910, 0.3020
0.6500, 1.0000, 0.3020, 0.0980, 0.2190, 0.1606, 0.4000, 0.5000, 0.4522, 0.0840, 0.1980
0.4100, 0.0000, 0.3240, 0.0940, 0.1710, 0.1044, 0.5600, 0.3000, 0.3970, 0.0760, 0.0950
0.5500, 1.0000, 0.2340, 0.0830, 0.1660, 0.1016, 0.4600, 0.4000, 0.4522, 0.0960, 0.0530
0.2200, 0.0000, 0.1930, 0.0820, 0.1560, 0.0932, 0.5200, 0.3000, 0.3989, 0.0710, 0.1340
0.5600, 0.0000, 0.3100, 0.0787, 0.1870, 0.1414, 0.3400, 0.5500, 0.4060, 0.0900, 0.1440
0.5400, 1.0000, 0.3060, 0.1033, 0.1440, 0.0798, 0.3000, 0.4800, 0.5142, 0.1010, 0.2320
0.5900, 1.0000, 0.2550, 0.0953, 0.1900, 0.1394, 0.3500, 0.5430, 0.4357, 0.1170, 0.0810
0.6000, 1.0000, 0.2340, 0.0880, 0.1530, 0.0898, 0.5800, 0.3000, 0.3258, 0.0950, 0.1040
0.5400, 0.0000, 0.2680, 0.0870, 0.2060, 0.1220, 0.6800, 0.3000, 0.4382, 0.0800, 0.0590
0.2500, 0.0000, 0.2830, 0.0870, 0.1930, 0.1280, 0.4900, 0.4000, 0.4382, 0.0920, 0.2460
0.5400, 1.0000, 0.2770, 0.1130, 0.2000, 0.1284, 0.3700, 0.5000, 0.5153, 0.1130, 0.2970
0.5500, 0.0000, 0.3660, 0.1130, 0.1990, 0.0944, 0.4300, 0.4630, 0.5730, 0.0970, 0.2580
0.4000, 1.0000, 0.2650, 0.0930, 0.2360, 0.1470, 0.3700, 0.7000, 0.5561, 0.0920, 0.2290
0.6200, 1.0000, 0.3180, 0.1150, 0.1990, 0.1286, 0.4400, 0.5000, 0.4883, 0.0980, 0.2750
0.6500, 0.0000, 0.2440, 0.1200, 0.2220, 0.1356, 0.3700, 0.6000, 0.5509, 0.1240, 0.2810
0.3300, 1.0000, 0.2540, 0.1020, 0.2060, 0.1410, 0.3900, 0.5000, 0.4868, 0.1050, 0.1790
0.5300, 0.0000, 0.2200, 0.0940, 0.1750, 0.0880, 0.5900, 0.3000, 0.4942, 0.0980, 0.2000
0.3500, 0.0000, 0.2680, 0.0980, 0.1620, 0.1036, 0.4500, 0.4000, 0.4205, 0.0860, 0.2000
0.6600, 0.0000, 0.2800, 0.1010, 0.1950, 0.1292, 0.4000, 0.5000, 0.4860, 0.0940, 0.1730
0.6200, 1.0000, 0.3390, 0.1010, 0.2210, 0.1564, 0.3500, 0.6000, 0.4997, 0.1030, 0.1800
0.5000, 1.0000, 0.2960, 0.0943, 0.3000, 0.2424, 0.3300, 0.9090, 0.4812, 0.1090, 0.0840
0.4700, 0.0000, 0.2860, 0.0970, 0.1640, 0.0906, 0.5600, 0.3000, 0.4466, 0.0880, 0.1210
0.4700, 1.0000, 0.2560, 0.0940, 0.1650, 0.0748, 0.4000, 0.4000, 0.5526, 0.0930, 0.1610
0.2400, 0.0000, 0.2070, 0.0870, 0.1490, 0.0806, 0.6100, 0.2000, 0.3611, 0.0780, 0.0990
0.5800, 1.0000, 0.2620, 0.0910, 0.2170, 0.1242, 0.7100, 0.3000, 0.4691, 0.0680, 0.1090
0.3400, 0.0000, 0.2060, 0.0870, 0.1850, 0.1122, 0.5800, 0.3000, 0.4304, 0.0740, 0.1150
0.5100, 0.0000, 0.2790, 0.0960, 0.1960, 0.1222, 0.4200, 0.5000, 0.5069, 0.1200, 0.2680
0.3100, 1.0000, 0.3530, 0.1250, 0.1870, 0.1124, 0.4800, 0.4000, 0.4890, 0.1090, 0.2740
0.2200, 0.0000, 0.1990, 0.0750, 0.1750, 0.1086, 0.5400, 0.3000, 0.4127, 0.0720, 0.1580
0.5300, 1.0000, 0.2440, 0.0920, 0.2140, 0.1460, 0.5000, 0.4000, 0.4500, 0.0970, 0.1070
0.3700, 1.0000, 0.2140, 0.0830, 0.1280, 0.0696, 0.4900, 0.3000, 0.3850, 0.0840, 0.0830
0.2800, 0.0000, 0.3040, 0.0850, 0.1980, 0.1156, 0.6700, 0.3000, 0.4344, 0.0800, 0.1030
0.4700, 0.0000, 0.3160, 0.0840, 0.1540, 0.0880, 0.3000, 0.5100, 0.5199, 0.1050, 0.2720
0.2300, 0.0000, 0.1880, 0.0780, 0.1450, 0.0720, 0.6300, 0.2000, 0.3912, 0.0860, 0.0850
0.5000, 0.0000, 0.3100, 0.1230, 0.1780, 0.1050, 0.4800, 0.4000, 0.4828, 0.0880, 0.2800
0.5800, 1.0000, 0.3670, 0.1170, 0.1660, 0.0938, 0.4400, 0.4000, 0.4949, 0.1090, 0.3360
0.5500, 0.0000, 0.3210, 0.1100, 0.1640, 0.0842, 0.4200, 0.4000, 0.5242, 0.0900, 0.2810
0.6000, 1.0000, 0.2770, 0.1070, 0.1670, 0.1146, 0.3800, 0.4000, 0.4277, 0.0950, 0.1180
0.4100, 0.0000, 0.3080, 0.0810, 0.2140, 0.1520, 0.2800, 0.7600, 0.5136, 0.1230, 0.3170
0.6000, 1.0000, 0.2750, 0.1060, 0.2290, 0.1438, 0.5100, 0.4000, 0.5142, 0.0910, 0.2350
0.4000, 0.0000, 0.2690, 0.0920, 0.2030, 0.1198, 0.7000, 0.3000, 0.4190, 0.0810, 0.0600
0.5700, 1.0000, 0.3070, 0.0900, 0.2040, 0.1478, 0.3400, 0.6000, 0.4710, 0.0930, 0.1740
0.3700, 0.0000, 0.3830, 0.1130, 0.1650, 0.0946, 0.5300, 0.3000, 0.4466, 0.0790, 0.2590
0.4000, 1.0000, 0.3190, 0.0950, 0.1980, 0.1356, 0.3800, 0.5000, 0.4804, 0.0930, 0.1780
0.3300, 0.0000, 0.3500, 0.0890, 0.2000, 0.1304, 0.4200, 0.4760, 0.4927, 0.1010, 0.1280
0.3200, 1.0000, 0.2780, 0.0890, 0.2160, 0.1462, 0.5500, 0.4000, 0.4304, 0.0910, 0.0960
0.3500, 1.0000, 0.2590, 0.0810, 0.1740, 0.1024, 0.3100, 0.6000, 0.5313, 0.0820, 0.1260
0.5500, 0.0000, 0.3290, 0.1020, 0.1640, 0.1062, 0.4100, 0.4000, 0.4431, 0.0890, 0.2880
0.4900, 0.0000, 0.2600, 0.0930, 0.1830, 0.1002, 0.6400, 0.3000, 0.4543, 0.0880, 0.0880
0.3900, 1.0000, 0.2630, 0.1150, 0.2180, 0.1582, 0.3200, 0.7000, 0.4935, 0.1090, 0.2920
0.6000, 1.0000, 0.2230, 0.1130, 0.1860, 0.1258, 0.4600, 0.4000, 0.4263, 0.0940, 0.0710
0.6700, 1.0000, 0.2830, 0.0930, 0.2040, 0.1322, 0.4900, 0.4000, 0.4736, 0.0920, 0.1970
0.4100, 1.0000, 0.3200, 0.1090, 0.2510, 0.1706, 0.4900, 0.5000, 0.5056, 0.1030, 0.1860
0.4400, 0.0000, 0.2540, 0.0950, 0.1620, 0.0926, 0.5300, 0.3000, 0.4407, 0.0830, 0.0250
0.4800, 1.0000, 0.2330, 0.0893, 0.2120, 0.1428, 0.4600, 0.4610, 0.4754, 0.0980, 0.0840
0.4500, 0.0000, 0.2030, 0.0743, 0.1900, 0.1262, 0.4900, 0.3880, 0.4304, 0.0790, 0.0960
0.4700, 0.0000, 0.3040, 0.1200, 0.1990, 0.1200, 0.4600, 0.4000, 0.5106, 0.0870, 0.1950
0.4600, 0.0000, 0.2060, 0.0730, 0.1720, 0.1070, 0.5100, 0.3000, 0.4249, 0.0800, 0.0530
0.3600, 1.0000, 0.3230, 0.1150, 0.2860, 0.1994, 0.3900, 0.7000, 0.5472, 0.1120, 0.2170
0.3400, 0.0000, 0.2920, 0.0730, 0.1720, 0.1082, 0.4900, 0.4000, 0.4304, 0.0910, 0.1720
0.5300, 1.0000, 0.3310, 0.1170, 0.1830, 0.1190, 0.4800, 0.4000, 0.4382, 0.1060, 0.1310
0.6100, 0.0000, 0.2460, 0.1010, 0.2090, 0.1068, 0.7700, 0.3000, 0.4836, 0.0880, 0.2140
0.3700, 0.0000, 0.2020, 0.0810, 0.1620, 0.0878, 0.6300, 0.3000, 0.4025, 0.0880, 0.0590
0.3300, 1.0000, 0.2080, 0.0840, 0.1250, 0.0702, 0.4600, 0.3000, 0.3784, 0.0660, 0.0700
0.6800, 0.0000, 0.3280, 0.1057, 0.2050, 0.1164, 0.4000, 0.5130, 0.5493, 0.1170, 0.2200
0.4900, 1.0000, 0.3190, 0.0940, 0.2340, 0.1558, 0.3400, 0.7000, 0.5398, 0.1220, 0.2680
0.4800, 0.0000, 0.2390, 0.1090, 0.2320, 0.1052, 0.3700, 0.6000, 0.6107, 0.0960, 0.1520
0.5500, 1.0000, 0.2450, 0.0840, 0.1790, 0.1058, 0.6600, 0.3000, 0.3584, 0.0870, 0.0470
0.4300, 0.0000, 0.2210, 0.0660, 0.1340, 0.0772, 0.4500, 0.3000, 0.4078, 0.0800, 0.0740
0.6000, 1.0000, 0.3300, 0.0970, 0.2170, 0.1256, 0.4500, 0.5000, 0.5447, 0.1120, 0.2950
0.3100, 1.0000, 0.1900, 0.0930, 0.1370, 0.0730, 0.4700, 0.3000, 0.4443, 0.0780, 0.1010
0.5300, 1.0000, 0.2730, 0.0820, 0.1190, 0.0550, 0.3900, 0.3000, 0.4828, 0.0930, 0.1510
0.6700, 0.0000, 0.2280, 0.0870, 0.1660, 0.0986, 0.5200, 0.3000, 0.4344, 0.0920, 0.1270
0.6100, 1.0000, 0.2820, 0.1060, 0.2040, 0.1320, 0.5200, 0.4000, 0.4605, 0.0960, 0.2370
0.6200, 0.0000, 0.2890, 0.0873, 0.2060, 0.1272, 0.3300, 0.6240, 0.5434, 0.0990, 0.2250
0.6000, 0.0000, 0.2560, 0.0870, 0.2070, 0.1258, 0.6900, 0.3000, 0.4111, 0.0840, 0.0810
0.4200, 0.0000, 0.2490, 0.0910, 0.2040, 0.1418, 0.3800, 0.5000, 0.4796, 0.0890, 0.1510
0.3800, 1.0000, 0.2680, 0.1050, 0.1810, 0.1192, 0.3700, 0.5000, 0.4820, 0.0910, 0.1070
0.6200, 0.0000, 0.2240, 0.0790, 0.2220, 0.1474, 0.5900, 0.4000, 0.4357, 0.0760, 0.0640
0.6100, 1.0000, 0.2690, 0.1110, 0.2360, 0.1724, 0.3900, 0.6000, 0.4812, 0.0890, 0.1380
0.6100, 1.0000, 0.2310, 0.1130, 0.1860, 0.1144, 0.4700, 0.4000, 0.4812, 0.1050, 0.1850
0.5300, 0.0000, 0.2860, 0.0880, 0.1710, 0.0988, 0.4100, 0.4000, 0.5050, 0.0990, 0.2650
0.2800, 1.0000, 0.2470, 0.0970, 0.1750, 0.0996, 0.3200, 0.5000, 0.5380, 0.0870, 0.1010
0.2600, 1.0000, 0.3030, 0.0890, 0.2180, 0.1522, 0.3100, 0.7000, 0.5159, 0.0820, 0.1370
0.3000, 0.0000, 0.2130, 0.0870, 0.1340, 0.0630, 0.6300, 0.2000, 0.3689, 0.0660, 0.1430
0.5000, 0.0000, 0.2610, 0.1090, 0.2430, 0.1606, 0.6200, 0.4000, 0.4625, 0.0890, 0.1410
0.4800, 0.0000, 0.2020, 0.0950, 0.1870, 0.1174, 0.5300, 0.4000, 0.4419, 0.0850, 0.0790
0.5100, 0.0000, 0.2520, 0.1030, 0.1760, 0.1122, 0.3700, 0.5000, 0.4898, 0.0900, 0.2920
0.4700, 1.0000, 0.2250, 0.0820, 0.1310, 0.0668, 0.4100, 0.3000, 0.4754, 0.0890, 0.1780
0.6400, 1.0000, 0.2350, 0.0970, 0.2030, 0.1290, 0.5900, 0.3000, 0.4318, 0.0770, 0.0910
0.5100, 1.0000, 0.2590, 0.0760, 0.2400, 0.1690, 0.3900, 0.6000, 0.5075, 0.0960, 0.1160
0.3000, 0.0000, 0.2090, 0.1040, 0.1520, 0.0838, 0.4700, 0.3000, 0.4663, 0.0970, 0.0860
0.5600, 1.0000, 0.2870, 0.0990, 0.2080, 0.1464, 0.3900, 0.5000, 0.4727, 0.0970, 0.1220
0.4200, 0.0000, 0.2210, 0.0850, 0.2130, 0.1386, 0.6000, 0.4000, 0.4277, 0.0940, 0.0720
0.6200, 1.0000, 0.2670, 0.1150, 0.1830, 0.1240, 0.3500, 0.5000, 0.4788, 0.1000, 0.1290
0.3400, 0.0000, 0.3140, 0.0870, 0.1490, 0.0938, 0.4600, 0.3000, 0.3829, 0.0770, 0.1420
0.6000, 0.0000, 0.2220, 0.1047, 0.2210, 0.1054, 0.6000, 0.3680, 0.5628, 0.0930, 0.0900
0.6400, 0.0000, 0.2100, 0.0923, 0.2270, 0.1468, 0.6500, 0.3490, 0.4331, 0.1020, 0.1580
0.3900, 1.0000, 0.2120, 0.0900, 0.1820, 0.1104, 0.6000, 0.3000, 0.4060, 0.0980, 0.0390
0.7100, 1.0000, 0.2650, 0.1050, 0.2810, 0.1736, 0.5500, 0.5000, 0.5568, 0.0840, 0.1960
0.4800, 1.0000, 0.2920, 0.1100, 0.2180, 0.1516, 0.3900, 0.6000, 0.4920, 0.0980, 0.2220
0.7900, 1.0000, 0.2700, 0.1030, 0.1690, 0.1108, 0.3700, 0.5000, 0.4663, 0.1100, 0.2770
0.4000, 0.0000, 0.3070, 0.0990, 0.1770, 0.0854, 0.5000, 0.4000, 0.5338, 0.0850, 0.0990
0.4900, 1.0000, 0.2880, 0.0920, 0.2070, 0.1400, 0.4400, 0.5000, 0.4745, 0.0920, 0.1960
0.5100, 0.0000, 0.3060, 0.1030, 0.1980, 0.1066, 0.5700, 0.3000, 0.5148, 0.1000, 0.2020
0.5700, 0.0000, 0.3010, 0.1170, 0.2020, 0.1396, 0.4200, 0.5000, 0.4625, 0.1200, 0.1550
0.5900, 1.0000, 0.2470, 0.1140, 0.1520, 0.1048, 0.2900, 0.5000, 0.4511, 0.0880, 0.0770
0.5100, 0.0000, 0.2770, 0.0990, 0.2290, 0.1456, 0.6900, 0.3000, 0.4277, 0.0770, 0.1910
0.7400, 0.0000, 0.2980, 0.1010, 0.1710, 0.1048, 0.5000, 0.3000, 0.4394, 0.0860, 0.0700
0.6700, 0.0000, 0.2670, 0.1050, 0.2250, 0.1354, 0.6900, 0.3000, 0.4635, 0.0960, 0.0730
0.4900, 0.0000, 0.1980, 0.0880, 0.1880, 0.1148, 0.5700, 0.3000, 0.4394, 0.0930, 0.0490
0.5700, 0.0000, 0.2330, 0.0880, 0.1550, 0.0636, 0.7800, 0.2000, 0.4205, 0.0780, 0.0650
0.5600, 1.0000, 0.3510, 0.1230, 0.1640, 0.0950, 0.3800, 0.4000, 0.5043, 0.1170, 0.2630
0.5200, 1.0000, 0.2970, 0.1090, 0.2280, 0.1628, 0.3100, 0.8000, 0.5142, 0.1030, 0.2480
0.6900, 0.0000, 0.2930, 0.1240, 0.2230, 0.1390, 0.5400, 0.4000, 0.5011, 0.1020, 0.2960
0.3700, 0.0000, 0.2030, 0.0830, 0.1850, 0.1246, 0.3800, 0.5000, 0.4719, 0.0880, 0.2140
0.2400, 0.0000, 0.2250, 0.0890, 0.1410, 0.0680, 0.5200, 0.3000, 0.4654, 0.0840, 0.1850
0.5500, 1.0000, 0.2270, 0.0930, 0.1540, 0.0942, 0.5300, 0.3000, 0.3526, 0.0750, 0.0780
0.3600, 0.0000, 0.2280, 0.0870, 0.1780, 0.1160, 0.4100, 0.4000, 0.4654, 0.0820, 0.0930
0.4200, 1.0000, 0.2400, 0.1070, 0.1500, 0.0850, 0.4400, 0.3000, 0.4654, 0.0960, 0.2520
0.2100, 0.0000, 0.2420, 0.0760, 0.1470, 0.0770, 0.5300, 0.3000, 0.4443, 0.0790, 0.1500
0.4100, 0.0000, 0.2020, 0.0620, 0.1530, 0.0890, 0.5000, 0.3000, 0.4249, 0.0890, 0.0770
0.5700, 1.0000, 0.2940, 0.1090, 0.1600, 0.0876, 0.3100, 0.5000, 0.5333, 0.0920, 0.2080
0.2000, 1.0000, 0.2210, 0.0870, 0.1710, 0.0996, 0.5800, 0.3000, 0.4205, 0.0780, 0.0770
0.6700, 1.0000, 0.2360, 0.1113, 0.1890, 0.1054, 0.7000, 0.2700, 0.4220, 0.0930, 0.1080
0.3400, 0.0000, 0.2520, 0.0770, 0.1890, 0.1206, 0.5300, 0.4000, 0.4344, 0.0790, 0.1600
0.4100, 1.0000, 0.2490, 0.0860, 0.1920, 0.1150, 0.6100, 0.3000, 0.4382, 0.0940, 0.0530
0.3800, 1.0000, 0.3300, 0.0780, 0.3010, 0.2150, 0.5000, 0.6020, 0.5193, 0.1080, 0.2200
0.5100, 0.0000, 0.2350, 0.1010, 0.1950, 0.1210, 0.5100, 0.4000, 0.4745, 0.0940, 0.1540
0.5200, 1.0000, 0.2640, 0.0913, 0.2180, 0.1520, 0.3900, 0.5590, 0.4905, 0.0990, 0.2590
0.6700, 0.0000, 0.2980, 0.0800, 0.1720, 0.0934, 0.6300, 0.3000, 0.4357, 0.0820, 0.0900
0.6100, 0.0000, 0.3000, 0.1080, 0.1940, 0.1000, 0.5200, 0.3730, 0.5347, 0.1050, 0.2460
0.6700, 1.0000, 0.2500, 0.1117, 0.1460, 0.0934, 0.3300, 0.4420, 0.4585, 0.1030, 0.1240
0.5600, 0.0000, 0.2700, 0.1050, 0.2470, 0.1606, 0.5400, 0.5000, 0.5088, 0.0940, 0.0670
0.6400, 0.0000, 0.2000, 0.0747, 0.1890, 0.1148, 0.6200, 0.3050, 0.4111, 0.0910, 0.0720
0.5800, 1.0000, 0.2550, 0.1120, 0.1630, 0.1106, 0.2900, 0.6000, 0.4762, 0.0860, 0.2570
0.5500, 0.0000, 0.2820, 0.0910, 0.2500, 0.1402, 0.6700, 0.4000, 0.5366, 0.1030, 0.2620
0.6200, 1.0000, 0.3330, 0.1140, 0.1820, 0.1140, 0.3800, 0.5000, 0.5011, 0.0960, 0.2750
0.5700, 1.0000, 0.2560, 0.0960, 0.2000, 0.1330, 0.5200, 0.3850, 0.4318, 0.1050, 0.1770
0.2000, 1.0000, 0.2420, 0.0880, 0.1260, 0.0722, 0.4500, 0.3000, 0.3784, 0.0740, 0.0710
0.5300, 1.0000, 0.2210, 0.0980, 0.1650, 0.1052, 0.4700, 0.4000, 0.4159, 0.0810, 0.0470
0.3200, 1.0000, 0.3140, 0.0890, 0.1530, 0.0842, 0.5600, 0.3000, 0.4159, 0.0900, 0.1870
0.4100, 0.0000, 0.2310, 0.0860, 0.1480, 0.0780, 0.5800, 0.3000, 0.4094, 0.0600, 0.1250
0.6000, 0.0000, 0.2340, 0.0767, 0.2470, 0.1480, 0.6500, 0.3800, 0.5136, 0.0770, 0.0780
0.2600, 0.0000, 0.1880, 0.0830, 0.1910, 0.1036, 0.6900, 0.3000, 0.4522, 0.0690, 0.0510
0.3700, 0.0000, 0.3080, 0.1120, 0.2820, 0.1972, 0.4300, 0.7000, 0.5342, 0.1010, 0.2580
0.4500, 0.0000, 0.3200, 0.1100, 0.2240, 0.1342, 0.4500, 0.5000, 0.5412, 0.0930, 0.2150
0.6700, 0.0000, 0.3160, 0.1160, 0.1790, 0.0904, 0.4100, 0.4000, 0.5472, 0.1000, 0.3030
0.3400, 1.0000, 0.3550, 0.1200, 0.2330, 0.1466, 0.3400, 0.7000, 0.5568, 0.1010, 0.2430
0.5000, 0.0000, 0.3190, 0.0783, 0.2070, 0.1492, 0.3800, 0.5450, 0.4595, 0.0840, 0.0910
0.7100, 0.0000, 0.2950, 0.0970, 0.2270, 0.1516, 0.4500, 0.5000, 0.5024, 0.1080, 0.1500
0.5700, 1.0000, 0.3160, 0.1170, 0.2250, 0.1076, 0.4000, 0.6000, 0.5958, 0.1130, 0.3100
0.4900, 0.0000, 0.2030, 0.0930, 0.1840, 0.1030, 0.6100, 0.3000, 0.4605, 0.0930, 0.1530
0.3500, 0.0000, 0.4130, 0.0810, 0.1680, 0.1028, 0.3700, 0.5000, 0.4949, 0.0940, 0.3460
0.4100, 1.0000, 0.2120, 0.1020, 0.1840, 0.1004, 0.6400, 0.3000, 0.4585, 0.0790, 0.0630
0.7000, 1.0000, 0.2410, 0.0823, 0.1940, 0.1492, 0.3100, 0.6260, 0.4234, 0.1050, 0.0890
0.5200, 0.0000, 0.2300, 0.1070, 0.1790, 0.1237, 0.4250, 0.4210, 0.4159, 0.0930, 0.0500
0.6000, 0.0000, 0.2560, 0.0780, 0.1950, 0.0954, 0.9100, 0.2000, 0.3761, 0.0870, 0.0390
0.6200, 0.0000, 0.2250, 0.1250, 0.2150, 0.0990, 0.9800, 0.2000, 0.4500, 0.0950, 0.1030
0.4400, 1.0000, 0.3820, 0.1230, 0.2010, 0.1266, 0.4400, 0.5000, 0.5024, 0.0920, 0.3080
0.2800, 1.0000, 0.1920, 0.0810, 0.1550, 0.0946, 0.5100, 0.3000, 0.3850, 0.0870, 0.1160
0.5800, 1.0000, 0.2900, 0.0850, 0.1560, 0.1092, 0.3600, 0.4000, 0.3989, 0.0860, 0.1450
0.3900, 1.0000, 0.2400, 0.0897, 0.1900, 0.1136, 0.5200, 0.3650, 0.4804, 0.1010, 0.0740
0.3400, 1.0000, 0.2060, 0.0980, 0.1830, 0.0920, 0.8300, 0.2000, 0.3689, 0.0920, 0.0450
0.6500, 0.0000, 0.2630, 0.0700, 0.2440, 0.1662, 0.5100, 0.5000, 0.4898, 0.0980, 0.1150
0.6600, 1.0000, 0.3460, 0.1150, 0.2040, 0.1394, 0.3600, 0.6000, 0.4963, 0.1090, 0.2640
0.5100, 0.0000, 0.2340, 0.0870, 0.2200, 0.1088, 0.9300, 0.2000, 0.4511, 0.0820, 0.0870
0.5000, 1.0000, 0.2920, 0.1190, 0.1620, 0.0852, 0.5400, 0.3000, 0.4736, 0.0950, 0.2020
0.5900, 1.0000, 0.2720, 0.1070, 0.1580, 0.1020, 0.3900, 0.4000, 0.4443, 0.0930, 0.1270
0.5200, 0.0000, 0.2700, 0.0783, 0.1340, 0.0730, 0.4400, 0.3050, 0.4443, 0.0690, 0.1820
0.6900, 1.0000, 0.2450, 0.1080, 0.2430, 0.1364, 0.4000, 0.6000, 0.5808, 0.1000, 0.2410
0.5300, 0.0000, 0.2410, 0.1050, 0.1840, 0.1134, 0.4600, 0.4000, 0.4812, 0.0950, 0.0660
0.4700, 1.0000, 0.2530, 0.0980, 0.1730, 0.1056, 0.4400, 0.4000, 0.4762, 0.1080, 0.0940
0.5200, 0.0000, 0.2880, 0.1130, 0.2800, 0.1740, 0.6700, 0.4000, 0.5273, 0.0860, 0.2830
0.3900, 0.0000, 0.2090, 0.0950, 0.1500, 0.0656, 0.6800, 0.2000, 0.4407, 0.0950, 0.0640
0.6700, 1.0000, 0.2300, 0.0700, 0.1840, 0.1280, 0.3500, 0.5000, 0.4654, 0.0990, 0.1020
0.5900, 1.0000, 0.2410, 0.0960, 0.1700, 0.0986, 0.5400, 0.3000, 0.4466, 0.0850, 0.2000
0.5100, 1.0000, 0.2810, 0.1060, 0.2020, 0.1222, 0.5500, 0.4000, 0.4820, 0.0870, 0.2650
0.2300, 1.0000, 0.1800, 0.0780, 0.1710, 0.0960, 0.4800, 0.4000, 0.4905, 0.0920, 0.0940
0.6800, 0.0000, 0.2590, 0.0930, 0.2530, 0.1812, 0.5300, 0.5000, 0.4543, 0.0980, 0.2300
0.4400, 0.0000, 0.2150, 0.0850, 0.1570, 0.0922, 0.5500, 0.3000, 0.3892, 0.0840, 0.1810
0.6000, 1.0000, 0.2430, 0.1030, 0.1410, 0.0866, 0.3300, 0.4000, 0.4673, 0.0780, 0.1560
0.5200, 0.0000, 0.2450, 0.0900, 0.1980, 0.1290, 0.2900, 0.7000, 0.5298, 0.0860, 0.2330
0.3800, 0.0000, 0.2130, 0.0720, 0.1650, 0.0602, 0.8800, 0.2000, 0.4431, 0.0900, 0.0600
0.6100, 0.0000, 0.2580, 0.0900, 0.2800, 0.1954, 0.5500, 0.5000, 0.4997, 0.0900, 0.2190
0.6800, 1.0000, 0.2480, 0.1010, 0.2210, 0.1514, 0.6000, 0.4000, 0.3871, 0.0870, 0.0800
0.2800, 1.0000, 0.3150, 0.0830, 0.2280, 0.1494, 0.3800, 0.6000, 0.5313, 0.0830, 0.0680
0.6500, 1.0000, 0.3350, 0.1020, 0.1900, 0.1262, 0.3500, 0.5000, 0.4970, 0.1020, 0.3320
0.6900, 0.0000, 0.2810, 0.1130, 0.2340, 0.1428, 0.5200, 0.4000, 0.5278, 0.0770, 0.2480
0.5100, 0.0000, 0.2430, 0.0853, 0.1530, 0.0716, 0.7100, 0.2150, 0.3951, 0.0820, 0.0840
0.2900, 0.0000, 0.3500, 0.0983, 0.2040, 0.1426, 0.5000, 0.4080, 0.4043, 0.0910, 0.2000
0.5500, 1.0000, 0.2350, 0.0930, 0.1770, 0.1268, 0.4100, 0.4000, 0.3829, 0.0830, 0.0550
0.3400, 1.0000, 0.3000, 0.0830, 0.1850, 0.1072, 0.5300, 0.3000, 0.4820, 0.0920, 0.0850
0.6700, 0.0000, 0.2070, 0.0830, 0.1700, 0.0998, 0.5900, 0.3000, 0.4025, 0.0770, 0.0890
0.4900, 0.0000, 0.2560, 0.0760, 0.1610, 0.0998, 0.5100, 0.3000, 0.3932, 0.0780, 0.0310
0.5500, 1.0000, 0.2290, 0.0810, 0.1230, 0.0672, 0.4100, 0.3000, 0.4304, 0.0880, 0.1290
0.5900, 1.0000, 0.2510, 0.0900, 0.1630, 0.1014, 0.4600, 0.4000, 0.4357, 0.0910, 0.0830
0.5300, 0.0000, 0.3320, 0.0827, 0.1860, 0.1068, 0.4600, 0.4040, 0.5112, 0.1020, 0.2750
0.4800, 1.0000, 0.2410, 0.1100, 0.2090, 0.1346, 0.5800, 0.4000, 0.4407, 0.1000, 0.0650
0.5200, 0.0000, 0.2950, 0.1043, 0.2110, 0.1328, 0.4900, 0.4310, 0.4984, 0.0980, 0.1980
0.6900, 0.0000, 0.2960, 0.1220, 0.2310, 0.1284, 0.5600, 0.4000, 0.5451, 0.0860, 0.2360
0.6000, 1.0000, 0.2280, 0.1100, 0.2450, 0.1898, 0.3900, 0.6000, 0.4394, 0.0880, 0.2530
0.4600, 1.0000, 0.2270, 0.0830, 0.1830, 0.1258, 0.3200, 0.6000, 0.4836, 0.0750, 0.1240
0.5100, 1.0000, 0.2620, 0.1010, 0.1610, 0.0996, 0.4800, 0.3000, 0.4205, 0.0880, 0.0440
0.6700, 1.0000, 0.2350, 0.0960, 0.2070, 0.1382, 0.4200, 0.5000, 0.4898, 0.1110, 0.1720
0.4900, 0.0000, 0.2210, 0.0850, 0.1360, 0.0634, 0.6200, 0.2190, 0.3970, 0.0720, 0.1140
0.4600, 1.0000, 0.2650, 0.0940, 0.2470, 0.1602, 0.5900, 0.4000, 0.4935, 0.1110, 0.1420
0.4700, 0.0000, 0.3240, 0.1050, 0.1880, 0.1250, 0.4600, 0.4090, 0.4443, 0.0990, 0.1090
0.7500, 0.0000, 0.3010, 0.0780, 0.2220, 0.1542, 0.4400, 0.5050, 0.4779, 0.0970, 0.1800
0.2800, 0.0000, 0.2420, 0.0930, 0.1740, 0.1064, 0.5400, 0.3000, 0.4220, 0.0840, 0.1440
0.6500, 1.0000, 0.3130, 0.1100, 0.2130, 0.1280, 0.4700, 0.5000, 0.5247, 0.0910, 0.1630
0.4200, 0.0000, 0.3010, 0.0910, 0.1820, 0.1148, 0.4900, 0.4000, 0.4511, 0.0820, 0.1470
0.5100, 0.0000, 0.2450, 0.0790, 0.2120, 0.1286, 0.6500, 0.3000, 0.4522, 0.0910, 0.0970
0.5300, 1.0000, 0.2770, 0.0950, 0.1900, 0.1018, 0.4100, 0.5000, 0.5464, 0.1010, 0.2200
0.5400, 0.0000, 0.2320, 0.1107, 0.2380, 0.1628, 0.4800, 0.4960, 0.4913, 0.1080, 0.1900
0.7300, 0.0000, 0.2700, 0.1020, 0.2110, 0.1210, 0.6700, 0.3000, 0.4745, 0.0990, 0.1090
0.5400, 0.0000, 0.2680, 0.1080, 0.1760, 0.0806, 0.6700, 0.3000, 0.4956, 0.1060, 0.1910
0.4200, 0.0000, 0.2920, 0.0930, 0.2490, 0.1742, 0.4500, 0.6000, 0.5004, 0.0920, 0.1220
0.7500, 0.0000, 0.3120, 0.1177, 0.2290, 0.1388, 0.2900, 0.7900, 0.5724, 0.1060, 0.2300
0.5500, 1.0000, 0.3210, 0.1127, 0.2070, 0.0924, 0.2500, 0.8280, 0.6105, 0.1110, 0.2420
0.6800, 1.0000, 0.2570, 0.1090, 0.2330, 0.1126, 0.3500, 0.7000, 0.6057, 0.1050, 0.2480
0.5700, 0.0000, 0.2690, 0.0980, 0.2460, 0.1652, 0.3800, 0.7000, 0.5366, 0.0960, 0.2490
0.4800, 0.0000, 0.3140, 0.0753, 0.2420, 0.1516, 0.3800, 0.6370, 0.5568, 0.1030, 0.1920
0.6100, 1.0000, 0.2560, 0.0850, 0.1840, 0.1162, 0.3900, 0.5000, 0.4970, 0.0980, 0.1310
0.6900, 0.0000, 0.3700, 0.1030, 0.2070, 0.1314, 0.5500, 0.4000, 0.4635, 0.0900, 0.2370
0.3800, 0.0000, 0.3260, 0.0770, 0.1680, 0.1006, 0.4700, 0.4000, 0.4625, 0.0960, 0.0780
0.4500, 1.0000, 0.2120, 0.0940, 0.1690, 0.0968, 0.5500, 0.3000, 0.4454, 0.1020, 0.1350
0.5100, 1.0000, 0.2920, 0.1070, 0.1870, 0.1390, 0.3200, 0.6000, 0.4382, 0.0950, 0.2440
0.7100, 1.0000, 0.2400, 0.0840, 0.1380, 0.0858, 0.3900, 0.4000, 0.4190, 0.0900, 0.1990
0.5700, 0.0000, 0.3610, 0.1170, 0.1810, 0.1082, 0.3400, 0.5000, 0.5268, 0.1000, 0.2700
0.5600, 1.0000, 0.2580, 0.1030, 0.1770, 0.1144, 0.3400, 0.5000, 0.4963, 0.0990, 0.1640
0.3200, 1.0000, 0.2200, 0.0880, 0.1370, 0.0786, 0.4800, 0.3000, 0.3951, 0.0780, 0.0720
0.5000, 0.0000, 0.2190, 0.0910, 0.1900, 0.1112, 0.6700, 0.3000, 0.4078, 0.0770, 0.0960
0.4300, 0.0000, 0.3430, 0.0840, 0.2560, 0.1726, 0.3300, 0.8000, 0.5529, 0.1040, 0.3060
0.5400, 1.0000, 0.2520, 0.1150, 0.1810, 0.1200, 0.3900, 0.5000, 0.4701, 0.0920, 0.0910
0.3100, 0.0000, 0.2330, 0.0850, 0.1900, 0.1308, 0.4300, 0.4000, 0.4394, 0.0770, 0.2140
0.5600, 0.0000, 0.2570, 0.0800, 0.2440, 0.1516, 0.5900, 0.4000, 0.5118, 0.0950, 0.0950
0.4400, 0.0000, 0.2510, 0.1330, 0.1820, 0.1130, 0.5500, 0.3000, 0.4249, 0.0840, 0.2160
0.5700, 1.0000, 0.3190, 0.1110, 0.1730, 0.1162, 0.4100, 0.4000, 0.4369, 0.0870, 0.2630

Test data:


# diabetes_norm_test_100.txt
#
0.6400, 1.0000, 0.2840, 0.1110, 0.1840, 0.1270, 0.4100, 0.4000, 0.4382, 0.0970, 0.1780
0.4300, 0.0000, 0.2810, 0.1210, 0.1920, 0.1210, 0.6000, 0.3000, 0.4007, 0.0930, 0.1130
0.1900, 0.0000, 0.2530, 0.0830, 0.2250, 0.1566, 0.4600, 0.5000, 0.4719, 0.0840, 0.2000
0.7100, 1.0000, 0.2610, 0.0850, 0.2200, 0.1524, 0.4700, 0.5000, 0.4635, 0.0910, 0.1390
0.5000, 1.0000, 0.2800, 0.1040, 0.2820, 0.1968, 0.4400, 0.6000, 0.5328, 0.0950, 0.1390
0.5900, 1.0000, 0.2360, 0.0730, 0.1800, 0.1074, 0.5100, 0.4000, 0.4682, 0.0840, 0.0880
0.5700, 0.0000, 0.2450, 0.0930, 0.1860, 0.0966, 0.7100, 0.3000, 0.4522, 0.0910, 0.1480
0.4900, 1.0000, 0.2100, 0.0820, 0.1190, 0.0854, 0.2300, 0.5000, 0.3970, 0.0740, 0.0880
0.4100, 1.0000, 0.3200, 0.1260, 0.1980, 0.1042, 0.4900, 0.4000, 0.5412, 0.1240, 0.2430
0.2500, 1.0000, 0.2260, 0.0850, 0.1300, 0.0710, 0.4800, 0.3000, 0.4007, 0.0810, 0.0710
0.5200, 1.0000, 0.1970, 0.0810, 0.1520, 0.0534, 0.8200, 0.2000, 0.4419, 0.0820, 0.0770
0.3400, 0.0000, 0.2120, 0.0840, 0.2540, 0.1134, 0.5200, 0.5000, 0.6094, 0.0920, 0.1090
0.4200, 1.0000, 0.3060, 0.1010, 0.2690, 0.1722, 0.5000, 0.5000, 0.5455, 0.1060, 0.2720
0.2800, 1.0000, 0.2550, 0.0990, 0.1620, 0.1016, 0.4600, 0.4000, 0.4277, 0.0940, 0.0600
0.4700, 1.0000, 0.2330, 0.0900, 0.1950, 0.1258, 0.5400, 0.4000, 0.4331, 0.0730, 0.0540
0.3200, 1.0000, 0.3100, 0.1000, 0.1770, 0.0962, 0.4500, 0.4000, 0.5187, 0.0770, 0.2210
0.4300, 0.0000, 0.1850, 0.0870, 0.1630, 0.0936, 0.6100, 0.2670, 0.3738, 0.0800, 0.0900
0.5900, 1.0000, 0.2690, 0.1040, 0.1940, 0.1266, 0.4300, 0.5000, 0.4804, 0.1060, 0.3110
0.5300, 0.0000, 0.2830, 0.1010, 0.1790, 0.1070, 0.4800, 0.4000, 0.4788, 0.1010, 0.2810
0.6000, 0.0000, 0.2570, 0.1030, 0.1580, 0.0846, 0.6400, 0.2000, 0.3850, 0.0970, 0.1820
0.5400, 1.0000, 0.3610, 0.1150, 0.1630, 0.0984, 0.4300, 0.4000, 0.4682, 0.1010, 0.3210
0.3500, 1.0000, 0.2410, 0.0947, 0.1550, 0.0974, 0.3200, 0.4840, 0.4852, 0.0940, 0.0580
0.4900, 1.0000, 0.2580, 0.0890, 0.1820, 0.1186, 0.3900, 0.5000, 0.4804, 0.1150, 0.2620
0.5800, 0.0000, 0.2280, 0.0910, 0.1960, 0.1188, 0.4800, 0.4000, 0.4984, 0.1150, 0.2060
0.3600, 1.0000, 0.3910, 0.0900, 0.2190, 0.1358, 0.3800, 0.6000, 0.5421, 0.1030, 0.2330
0.4600, 1.0000, 0.4220, 0.0990, 0.2110, 0.1370, 0.4400, 0.5000, 0.5011, 0.0990, 0.2420
0.4400, 1.0000, 0.2660, 0.0990, 0.2050, 0.1090, 0.4300, 0.5000, 0.5580, 0.1110, 0.1230
0.4600, 0.0000, 0.2990, 0.0830, 0.1710, 0.1130, 0.3800, 0.4500, 0.4585, 0.0980, 0.1670
0.5400, 0.0000, 0.2100, 0.0780, 0.1880, 0.1074, 0.7000, 0.3000, 0.3970, 0.0730, 0.0630
0.6300, 1.0000, 0.2550, 0.1090, 0.2260, 0.1032, 0.4600, 0.5000, 0.5951, 0.0870, 0.1970
0.4100, 1.0000, 0.2420, 0.0900, 0.1990, 0.1236, 0.5700, 0.4000, 0.4522, 0.0860, 0.0710
0.2800, 0.0000, 0.2540, 0.0930, 0.1410, 0.0790, 0.4900, 0.3000, 0.4174, 0.0910, 0.1680
0.1900, 0.0000, 0.2320, 0.0750, 0.1430, 0.0704, 0.5200, 0.3000, 0.4635, 0.0720, 0.1400
0.6100, 1.0000, 0.2610, 0.1260, 0.2150, 0.1298, 0.5700, 0.4000, 0.4949, 0.0960, 0.2170
0.4800, 0.0000, 0.3270, 0.0930, 0.2760, 0.1986, 0.4300, 0.6420, 0.5148, 0.0910, 0.1210
0.5400, 1.0000, 0.2730, 0.1000, 0.2000, 0.1440, 0.3300, 0.6000, 0.4745, 0.0760, 0.2350
0.5300, 1.0000, 0.2660, 0.0930, 0.1850, 0.1224, 0.3600, 0.5000, 0.4890, 0.0820, 0.2450
0.4800, 0.0000, 0.2280, 0.1010, 0.1100, 0.0416, 0.5600, 0.2000, 0.4127, 0.0970, 0.0400
0.5300, 0.0000, 0.2880, 0.1117, 0.1450, 0.0872, 0.4600, 0.3150, 0.4078, 0.0850, 0.0520
0.2900, 1.0000, 0.1810, 0.0730, 0.1580, 0.0990, 0.4100, 0.4000, 0.4500, 0.0780, 0.1040
0.6200, 0.0000, 0.3200, 0.0880, 0.1720, 0.0690, 0.3800, 0.4000, 0.5784, 0.1000, 0.1320
0.5000, 1.0000, 0.2370, 0.0920, 0.1660, 0.0970, 0.5200, 0.3000, 0.4443, 0.0930, 0.0880
0.5800, 1.0000, 0.2360, 0.0960, 0.2570, 0.1710, 0.5900, 0.4000, 0.4905, 0.0820, 0.0690
0.5500, 1.0000, 0.2460, 0.1090, 0.1430, 0.0764, 0.5100, 0.3000, 0.4357, 0.0880, 0.2190
0.5400, 0.0000, 0.2260, 0.0900, 0.1830, 0.1042, 0.6400, 0.3000, 0.4304, 0.0920, 0.0720
0.3600, 0.0000, 0.2780, 0.0730, 0.1530, 0.1044, 0.4200, 0.4000, 0.3497, 0.0730, 0.2010
0.6300, 1.0000, 0.2410, 0.1110, 0.1840, 0.1122, 0.4400, 0.4000, 0.4935, 0.0820, 0.1100
0.4700, 1.0000, 0.2650, 0.0700, 0.1810, 0.1048, 0.6300, 0.3000, 0.4190, 0.0700, 0.0510
0.5100, 1.0000, 0.3280, 0.1120, 0.2020, 0.1006, 0.3700, 0.5000, 0.5775, 0.1090, 0.2770
0.4200, 0.0000, 0.1990, 0.0760, 0.1460, 0.0832, 0.5500, 0.3000, 0.3664, 0.0790, 0.0630
0.3700, 1.0000, 0.2360, 0.0940, 0.2050, 0.1388, 0.5300, 0.4000, 0.4190, 0.1070, 0.1180
0.2800, 0.0000, 0.2210, 0.0820, 0.1680, 0.1006, 0.5400, 0.3000, 0.4205, 0.0860, 0.0690
0.5800, 0.0000, 0.2810, 0.1110, 0.1980, 0.0806, 0.3100, 0.6000, 0.6068, 0.0930, 0.2730
0.3200, 0.0000, 0.2650, 0.0860, 0.1840, 0.1016, 0.5300, 0.4000, 0.4990, 0.0780, 0.2580
0.2500, 1.0000, 0.2350, 0.0880, 0.1430, 0.0808, 0.5500, 0.3000, 0.3584, 0.0830, 0.0430
0.6300, 0.0000, 0.2600, 0.0857, 0.1550, 0.0782, 0.4600, 0.3370, 0.5037, 0.0970, 0.1980
0.5200, 0.0000, 0.2780, 0.0850, 0.2190, 0.1360, 0.4900, 0.4000, 0.5136, 0.0750, 0.2420
0.6500, 1.0000, 0.2850, 0.1090, 0.2010, 0.1230, 0.4600, 0.4000, 0.5075, 0.0960, 0.2320
0.4200, 0.0000, 0.3060, 0.1210, 0.1760, 0.0928, 0.6900, 0.3000, 0.4263, 0.0890, 0.1750
0.5300, 0.0000, 0.2220, 0.0780, 0.1640, 0.0810, 0.7000, 0.2000, 0.4174, 0.1010, 0.0930
0.7900, 1.0000, 0.2330, 0.0880, 0.1860, 0.1284, 0.3300, 0.6000, 0.4812, 0.1020, 0.1680
0.4300, 0.0000, 0.3540, 0.0930, 0.1850, 0.1002, 0.4400, 0.4000, 0.5318, 0.1010, 0.2750
0.4400, 0.0000, 0.3140, 0.1150, 0.1650, 0.0976, 0.5200, 0.3000, 0.4344, 0.0890, 0.2930
0.6200, 1.0000, 0.3780, 0.1190, 0.1130, 0.0510, 0.3100, 0.4000, 0.5043, 0.0840, 0.2810
0.3300, 0.0000, 0.1890, 0.0700, 0.1620, 0.0918, 0.5900, 0.3000, 0.4025, 0.0580, 0.0720
0.5600, 0.0000, 0.3500, 0.0793, 0.1950, 0.1408, 0.4200, 0.4640, 0.4111, 0.0960, 0.1400
0.6600, 0.0000, 0.2170, 0.1260, 0.2120, 0.1278, 0.4500, 0.4710, 0.5278, 0.1010, 0.1890
0.3400, 1.0000, 0.2530, 0.1110, 0.2300, 0.1620, 0.3900, 0.6000, 0.4977, 0.0900, 0.1810
0.4600, 1.0000, 0.2380, 0.0970, 0.2240, 0.1392, 0.4200, 0.5000, 0.5366, 0.0810, 0.2090
0.5000, 0.0000, 0.3180, 0.0820, 0.1360, 0.0692, 0.5500, 0.2000, 0.4078, 0.0850, 0.1360
0.6900, 0.0000, 0.3430, 0.1130, 0.2000, 0.1238, 0.5400, 0.4000, 0.4710, 0.1120, 0.2610
0.3400, 0.0000, 0.2630, 0.0870, 0.1970, 0.1200, 0.6300, 0.3000, 0.4249, 0.0960, 0.1130
0.7100, 1.0000, 0.2700, 0.0933, 0.2690, 0.1902, 0.4100, 0.6560, 0.5242, 0.0930, 0.1310
0.4700, 0.0000, 0.2720, 0.0800, 0.2080, 0.1456, 0.3800, 0.6000, 0.4804, 0.0920, 0.1740
0.4100, 0.0000, 0.3380, 0.1233, 0.1870, 0.1270, 0.4500, 0.4160, 0.4318, 0.1000, 0.2570
0.3400, 0.0000, 0.3300, 0.0730, 0.1780, 0.1146, 0.5100, 0.3490, 0.4127, 0.0920, 0.0550
0.5100, 0.0000, 0.2410, 0.0870, 0.2610, 0.1756, 0.6900, 0.4000, 0.4407, 0.0930, 0.0840
0.4300, 0.0000, 0.2130, 0.0790, 0.1410, 0.0788, 0.5300, 0.3000, 0.3829, 0.0900, 0.0420
0.5500, 0.0000, 0.2300, 0.0947, 0.1900, 0.1376, 0.3800, 0.5000, 0.4277, 0.1060, 0.1460
0.5900, 1.0000, 0.2790, 0.1010, 0.2180, 0.1442, 0.3800, 0.6000, 0.5187, 0.0950, 0.2120
0.2700, 1.0000, 0.3360, 0.1100, 0.2460, 0.1566, 0.5700, 0.4000, 0.5088, 0.0890, 0.2330
0.5100, 1.0000, 0.2270, 0.1030, 0.2170, 0.1624, 0.3000, 0.7000, 0.4812, 0.0800, 0.0910
0.4900, 1.0000, 0.2740, 0.0890, 0.1770, 0.1130, 0.3700, 0.5000, 0.4905, 0.0970, 0.1110
0.2700, 0.0000, 0.2260, 0.0710, 0.1160, 0.0434, 0.5600, 0.2000, 0.4419, 0.0790, 0.1520
0.5700, 1.0000, 0.2320, 0.1073, 0.2310, 0.1594, 0.4100, 0.5630, 0.5030, 0.1120, 0.1200
0.3900, 1.0000, 0.2690, 0.0930, 0.1360, 0.0754, 0.4800, 0.3000, 0.4143, 0.0990, 0.0670
0.6200, 1.0000, 0.3460, 0.1200, 0.2150, 0.1292, 0.4300, 0.5000, 0.5366, 0.1230, 0.3100
0.3700, 0.0000, 0.2330, 0.0880, 0.2230, 0.1420, 0.6500, 0.3400, 0.4357, 0.0820, 0.0940
0.4600, 0.0000, 0.2110, 0.0800, 0.2050, 0.1444, 0.4200, 0.5000, 0.4533, 0.0870, 0.1830
0.6800, 1.0000, 0.2350, 0.1010, 0.1620, 0.0854, 0.5900, 0.3000, 0.4477, 0.0910, 0.0660
0.5100, 0.0000, 0.3150, 0.0930, 0.2310, 0.1440, 0.4900, 0.4700, 0.5252, 0.1170, 0.1730
0.4100, 0.0000, 0.2080, 0.0860, 0.2230, 0.1282, 0.8300, 0.3000, 0.4078, 0.0890, 0.0720
0.5300, 0.0000, 0.2650, 0.0970, 0.1930, 0.1224, 0.5800, 0.3000, 0.4143, 0.0990, 0.0490
0.4500, 0.0000, 0.2420, 0.0830, 0.1770, 0.1184, 0.4500, 0.4000, 0.4220, 0.0820, 0.0640
0.3300, 0.0000, 0.1950, 0.0800, 0.1710, 0.0854, 0.7500, 0.2000, 0.3970, 0.0800, 0.0480
0.6000, 1.0000, 0.2820, 0.1120, 0.1850, 0.1138, 0.4200, 0.4000, 0.4984, 0.0930, 0.1780
0.4700, 1.0000, 0.2490, 0.0750, 0.2250, 0.1660, 0.4200, 0.5000, 0.4443, 0.1020, 0.1040
0.6000, 1.0000, 0.2490, 0.0997, 0.1620, 0.1066, 0.4300, 0.3770, 0.4127, 0.0950, 0.1320
0.3600, 0.0000, 0.3000, 0.0950, 0.2010, 0.1252, 0.4200, 0.4790, 0.5130, 0.0850, 0.2200
0.3600, 0.0000, 0.1960, 0.0710, 0.2500, 0.1332, 0.9700, 0.3000, 0.4595, 0.0920, 0.0570

Posted in Machine Learning, Scikit | Leave a comment

Machine Learning Regression Baseline Accuracy

Posted on June 2, 2026 by jamesdmccaffrey

The goal of a machine learning regression model is to predict a single numeric value. For example, you might want to predict the bank account balance of a person based on their age, height, and annual income.

After a regression model has been created, it’s a good idea to evaluate the model. Common metrics are prediction percent accuracy, mean squared error or root mean squared error, and R2 (coefficient of determination). It’s also a good idea to compute a baseline accuracy and a baseline mean squared error, where the baseline metric assumes that you just predict the average of the target y values in the training data. If a prediction model has worse prediction accuracy than the baseline accuracy, something is seriously wrong with the model.

I put together a demo to illustrate the idea. I used linear regression, implemented from scratch using the C# language. The output of my demo is:

Begin C# linear regression baseline accuracy demo

Loading synthetic train (200) and test (40) data
Done

First three train X:
 -0.1660  0.4406 -0.9998 -0.3953 -0.7065
  0.0776 -0.1616  0.3704 -0.5911  0.7562
 -0.9452  0.3409 -0.1654  0.1174 -0.7192

First three train y:
  0.4840
  0.1568
  0.8054

Creating and training  model using left pseudo-inverse
Done

Model weights:
-0.2656  0.0333  -0.0454  0.0358  -0.1146
Bias/constant: 0.3619

Evaluating trained model

Accuracy train (within 0.10) = 0.4600
Accuracy test (within 0.10) = 0.6500

Baseline accuracy train (within 0.10) = 0.1100
Baseline accuracy test (within 0.10) = 0.0750

End demo

The demo data is synthetic. There are 5 predictor values. The trained model scores 46% accuracy on the 200-item training data and 65% accuracy on the 40-item test data. A prediction is scored correct if it is within 10% of the true target value. The model accuracy is poor because the synthetic data has a complex, non-linear structure.

But the baseline accuracy, where you just predict the average target y value for every input, is only 11% on the training data and 7.5% on the test data. Therefore, the linear regression model is much better than naively predicting the average y value.

In addition to computing model accuracy and baseline accuracy, it’s a good idea to compute model mean squared error and baseline mean squared error. And computing model R2 is a good idea too — R2 combines accuracy and baseline accuracy into one metric in the sense that R2 is accuracy relative to always guessing the mean target y value.

Galaxy Science Fiction was published from 1950 to 1980. It was the leading science fiction magazine of its time. The magazine had beautiful cover art. Famous space artist Chesley Bonestell is known for his accuracy in depicting technologies and space geology. Bonestell did three early covers in 1951: February (issue #5), May (#8), and November (#14).

Demo program. Replace “lt” (less than), “gt”, “lte”, “gte” with Boolean operator symbols (my blog editor often chokes on symbols).

using System;
using System.IO;
using System.Collections.Generic;

namespace LinearRegressionBaselines
{
  internal class LinearRegressionProgram
  {
    static void Main(string[] args)
    {
      Console.WriteLine("\nBegin C# linear regression" +
        " baseline accuracy demo ");

      // 1. load data
      Console.WriteLine("\nLoading synthetic train" +
        " (200) and test (40) data");
      string trainFile =
        "..\\..\\..\\Data\\synthetic_train_200.txt";
      int[] colsX = new int[] { 0, 1, 2, 3, 4 };
      double[][] trainX =
        MatLoad(trainFile, colsX, ',', "#");
      double[] trainY =
        MatToVec(MatLoad(trainFile,
        new int[] { 5 }, ',', "#"));

      string testFile =
        "..\\..\\..\\Data\\synthetic_test_40.txt";
      double[][] testX =
        MatLoad(testFile, colsX, ',', "#");
      double[] testY =
        MatToVec(MatLoad(testFile,
        new int[] { 5 }, ',', "#"));
      Console.WriteLine("Done ");

      Console.WriteLine("\nFirst three train X: ");
      for (int i = 0; i "lt" 3; ++i)
        VecShow(trainX[i], 4, 8);

      Console.WriteLine("\nFirst three train y: ");
      for (int i = 0; i "lt" 3; ++i)
        Console.WriteLine(trainY[i].ToString("F4").
          PadLeft(8));

      // 2. create and train model
      Console.WriteLine("\nCreating and training" +
       "  model using left pseudo-inverse ");

      LinearRegressor model = new LinearRegressor();
      model.TrainLeftPinv(trainX, trainY);
      Console.WriteLine("Done ");

      // 2b.show model parameters
      Console.WriteLine("\nModel weights: ");
      for (int i = 0; i "lt" model.weights.Length; ++i)
        Console.Write(model.weights[i].ToString("F4") + "  ");
      Console.WriteLine("\nBias/constant: " +
        model.bias.ToString("F4"));

      // 3. evaluate model
      Console.WriteLine("\nEvaluating trained model ");

      double accTrain = Eval.Accuracy(model,
        trainX, trainY, 0.10);
      Console.WriteLine("\nAccuracy train (within 0.10) = " +
        accTrain.ToString("F4"));
      double accTest = Eval.Accuracy(model,
        testX, testY, 0.10);
      Console.WriteLine("Accuracy test (within 0.10) = " +
        accTest.ToString("F4"));

      double baseAccTrain =
        Eval.BaselineAccuracy(trainX, trainY, 0.10);
      Console.WriteLine("\nBaseline accuracy train" +
        " (within 0.10) = " + baseAccTrain.ToString("F4"));
      double baseAccTest =
        Eval.BaselineAccuracy(testX, testY, 0.10);
      Console.WriteLine("Baseline accuracy test " +
        "(within 0.10) = " + baseAccTest.ToString("F4"));

      Console.WriteLine("\nEnd demo ");
      Console.ReadLine();
    } // Main

    // ------------------------------------------------------
    // helpers for Main()
    // ------------------------------------------------------

    static double[][] MatLoad(string fn, int[] usecols,
      char sep, string comment)
    {
      List"lt"double[]"gt" result = 
        new List"lt"double[]"gt"();
      string line = "";
      FileStream ifs = new FileStream(fn, FileMode.Open);
      StreamReader sr = new StreamReader(ifs);
      while ((line = sr.ReadLine()) != null)
      {
        if (line.StartsWith(comment) == true)
          continue;
        string[] tokens = line.Split(sep);
        List"lt"double"gt" lst = new List"lt"double"gt"();
        for (int j = 0; j "lt" usecols.Length; ++j)
          lst.Add(double.Parse(tokens[usecols[j]]));
        double[] row = lst.ToArray();
        result.Add(row);
      }
      sr.Close(); ifs.Close();
      return result.ToArray();
    }

    static double[] MatToVec(double[][] M)
    {
      int nRows = M.Length;
      int nCols = M[0].Length;
      double[] result = new double[nRows * nCols];
      int k = 0;
      for (int i = 0; i "lt" nRows; ++i)
        for (int j = 0; j "lt" nCols; ++j)
          result[k++] = M[i][j];
      return result;
    }

    static void VecShow(double[] vec, int dec, int wid)
    {
      for (int i = 0; i "lt" vec.Length; ++i)
        Console.Write(vec[i].ToString("F" + dec).
          PadLeft(wid));
      Console.WriteLine("");
    }
  } // class Program

  // ========================================================

  public class LinearRegressor
  {
    public double[] weights;
    public double bias;
    private Random rnd;

    public LinearRegressor(int seed = 0)
    {
      this.weights = new double[0]; // keep compiler happy
      this.bias = 0;
      this.rnd = new Random(seed); // not used
    }

    // ------------------------------------------------------

    public void TrainLeftPinv(double[][] trainX,
      double[] trainY)
    {
      // pseudo-inverse via normal equations
      // wts_bias = (inv(Xt * X) * Xt) * trainY
      int dim = trainX[0].Length;
      this.weights = new double[dim];

      double[][] X = MatDesign(trainX);
      double[][] Xinv = Cholesky.MatPseudoInv(X);
      double[] biasAndWts =
        MatVecProduct(Xinv, trainY);

      // extract bias and weights
      this.bias = biasAndWts[0];
      for (int i = 1; i "lt" biasAndWts.Length; ++i)
        this.weights[i - 1] = biasAndWts[i];
      return;  // all done
    } // TrainClosed()

    private static double[] MatVecProduct(double[][] A,
        double[] v)
    {
      // helper for TrainLeftPinv()
      double[] result = new double[A.Length];
      for (int i = 0; i "lt" A.Length; ++i)
        for (int k = 0; k "lt" A[0].Length; ++k)
          result[i] += A[i][k] * v[k];
      return result;
    }

    // ------------------------------------------------------

    private static double[][] MatDesign(double[][] M)
    {
      // helper for TrainLeftPinv()
      int nRows = M.Length; int nCols = M[0].Length;
      double[][] result = new double[nRows][];
      for (int i = 0; i "lt" nRows; ++i)
        result[i] = new double[nCols + 1];
      for (int i = 0; i "lt" nRows; ++i)
      {
        result[i][0] = 1.0;
        for (int j = 1; j "lt" nCols + 1; ++j)
          result[i][j] = M[i][j - 1];
      }
      return result;
    }

    // ------------------------------------------------------

    public double Predict(double[] x)
    {
      double result = 0.0;
      for (int j = 0; j "lt" x.Length; ++j)
        result += x[j] * this.weights[j];
      result += this.bias;
      return result;
    }

    // ------------------------------------------------------

  } // class LinearRegressor

  // ========================================================

  public class Cholesky
  {
    // container class for MatPseudoInv() for TrainLeftPinv()

    public static double[][] MatPseudoInv(double[][] A)
    {
      // left pseudo-inverse via normal equations
      // nRows must be gte nCols
      // inv(At * A) * A
      double[][] At = MatTranspose(A);
      double[][] AtA = MatProduct(At, A);
      for (int i = 0; i "lt" AtA.Length; ++i)
        AtA[i][i] += 1.0e-8; /// condition before inv
      double[][] AtAinv = MatInvCholesky(AtA);
      double[][] pinv = MatProduct(AtAinv, At);
      return pinv;
    } // MatPseudoInv()

    // ------------------------------------------------------

    private static double[][] MatInvCholesky(double[][] A)
    {
      // A must be square, symmetric, positive definite
      int m = A.Length; int n = A[0].Length;  // m == n
      // 1. decompose A to L
      double[][] L = new double[n][];
      for (int i = 0; i "lt" n; ++i)
        L[i] = new double[n];

      for (int i = 0; i "lt" n; ++i)
      {
        for (int j = 0; j "lte" i; ++j)
        {
          double sum = 0.0;
          for (int k = 0; k "lt" j; ++k)
            sum += L[i][k] * L[j][k];
          if (i == j)
          {
            double tmp = A[i][i] - sum;
            if (tmp "lt" 0.0)
              throw new
                Exception("decomp Cholesky fatal");
            L[i][j] = Math.Sqrt(tmp);
          }
          else
          {
            if (L[j][j] == 0.0)
              throw new
                Exception("decomp Cholesky fatal ");
            L[i][j] = (A[i][j] - sum) / L[j][j];
          }
        } // j
      } // i

      // 2. compute inverse from L
      double[][] result = new double[n][];  // make Identity
      for (int i = 0; i "lt" n; ++i)
        result[i] = new double[n];
      for (int i = 0; i "lt" n; ++i)
        result[i][i] = 1.0;

      for (int k = 0; k "lt" n; ++k)
      {
        for (int j = 0; j "lt" n; j++)
        {
          for (int i = 0; i "lt" k; i++)
          {
            result[k][j] -= result[i][j] * L[k][i];
          }
          result[k][j] /= L[k][k];
        }
      }

      for (int k = n - 1; k "gte" 0; --k)
      {
        for (int j = 0; j "lt" n; j++)
        {
          for (int i = k + 1; i "lt" n; i++)
          {
            result[k][j] -= result[i][j] * L[i][k];
          }
          result[k][j] /= L[k][k];
        }
      }
      return result;
    } // MatInvCholesky()

    // ------------------------------------------------------

    private static double[][] MatTranspose(double[][] M)
    {
      int nr = M.Length; int nc = M[0].Length;
      double[][] result = new double[nc][]; // note
      for (int i = 0; i "lt" nc; ++i)
        result[i] = new double[nr];
      for (int i = 0; i "lt" nr; ++i)
        for (int j = 0; j "lt" nc; ++j)
          result[j][i] = M[i][j]; // note
      return result;
    }

    // ````````````````````````````````````````````````````

    private static double[][] MatProduct(double[][] A,
      double[][] B)
    {
      int aRows = A.Length; int aCols = A[0].Length;
      int bRows = B.Length; int bCols = B[0].Length;
      if (aCols != bRows)
        throw new Exception("Non-conformable matrices");

      double[][] result = new double[aRows][];
      for (int i = 0; i "lt" aRows; ++i)
        result[i] = new double[bCols];

      for (int i = 0; i "lt" aRows; ++i) // each row of A
        for (int j = 0; j "lt" bCols; ++j) // each col of B
          for (int k = 0; k "lt" aCols; ++k)
            result[i][j] += A[i][k] * B[k][j];
      return result;
    }

    // ------------------------------------------------------

  } // class Cholesky

  // ========================================================

  public class Eval
  {
    // uses 'dynamic' keyword (C# 4.0, 2010 and later)

    public static double Accuracy(dynamic model,
      double[][] dataX, double[] dataY, double pctClose)
    {
      int numCorrect = 0; int numWrong = 0;
      for (int i = 0; i "lt" dataX.Length; ++i)
      {
        double actualY = dataY[i];
        double predY = model.Predict(dataX[i]); // assume
        if (Math.Abs(predY - actualY) "lt"
          Math.Abs(pctClose * actualY))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / (numWrong + numCorrect);
    }

    // ------------------------------------------------------

    public static double BaselineAccuracy(double[][] dataX,
      double[] dataY, double pctClose)
    {
      // compute avg of y values
      int n = dataY.Length;
      double sum = 0.0;

      for (int i = 0; i "lt" n; ++i)
        sum += dataY[i];
      double meanY = sum / n;

      // always predict the mean
      int numCorrect = 0; int numWrong = 0;
      for (int i = 0; i "lt" dataX.Length; ++i)
      {
        double actualY = dataY[i];
        double predY = meanY;
        if (Math.Abs(predY - actualY) "lt"
          Math.Abs(pctClose * actualY))
          ++numCorrect;
        else
          ++numWrong;
      }
      return (numCorrect * 1.0) / (numWrong + numCorrect);
    }

    // ------------------------------------------------------

  } // class Eval


  // ========================================================

} // ns

Training data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning | Leave a comment

Trimmed Kernel Ridge Regression to Approximate Support Vector Regression Using Scikit

Posted on June 1, 2026 by jamesdmccaffrey

Two closely related machine learning regression techniques are kernel ridge regression (KRR) and support vector regression (SVR). Both techniques use a kernel function (usually the radial basis function, RBF) to compare two data items for similarity. Both techniques must store training data in order to make predictions, but KRR stores all training items, while SVR eliminates some of the items, leaving just the “support vectors” that need to be stored.

KRR often gives slightly better prediction accuracy than SVR, is easier to train than SVR, can handle very large datasets (KRR can use the SGD training technique), and is much, much easier to implement from scratch than SVR (SVR must use the very ugly quadratic programming technique).

One morning before work, I got the idea of combining KRR and SVR. Briefly, I train a KRR model as normal using all training data. Then I identify training data items that are predicted “too well” and remove them, leaving just pseudo support vectors. The SVR idea is that items that are predicted too well don’t help model accuracy very much. Then I retrain a new KRR model using only the reduced training data. This gives a trimmed/sparse KRR model that approximates an SVR model.

A trimmed KRR model and a SVR model won’t be exactly equivalent because KRR minimizes mean squared error and SVR minimizes a weird loss function called epsilon-insensitive loss. And trimmed KRR removes some of the stored training items after training, while SVR removes training items during training. But the two techniques should give nearly the same model in terms of predictive accuracy.

In my mind, this technique gives the advantages of KRR (ability to handle very large datasets) and the advantages of SVR (fewer stored model items and weights than KRR. . . if the technique works. Bottom line: it worked well.

I put together a demo using the scikit KernelRidge module. And the sparse KRR idea worked quite nicely. The output of the demo is:

Begin sparse KRR approximation to SVR demo

Loading synthetic train (200) and test (40) data
Done

First three train X:
[-0.1660  0.4406 -0.9998 -0.3953 -0.7065]
[ 0.0776 -0.1616  0.3704 -0.5911  0.7562]
[-0.9452  0.3409 -0.1654  0.1174 -0.7192]

First three train y:
0.4840
0.1568
0.8054

Creating preliminary KRR RBF model
Setting gamma = 0.1000
Setting alpha = 0.0010
Done

Training KRR model
Done

Removing non-pseudo-support vectors
Using epsilon = 0.003000
Number of pseudo support vectors = 138

Re-training sparse KRR model
Done

Evaluating model

Accuracy (within 0.10) train = 0.9710
Accuracy (within 0.10) test = 0.9500

MSE train = 0.0001
MSE test = 0.0002

The demo data is synthetic demo. There are 200 training items and 40 test items.

After training on 200 data items, those items that were predicted with a small error of less than 0.004850 were removed. There were 93 such well-predicted items, leaving a reduced training dataset of 107 items. After training a new model on the reduced dataset, the sparse model scored 97.10% accuracy on the reduced training data (134 out of 138 correct) and 95.00% accuracy on the test data (38 out of 40 correct).

An optimized scikit SVR model created 90 support vectors (better than the sparse KRR 138 items), with 96.50% accuracy on the training data (not quite as good as the sparse KRR model), and 95.00% accuracy on the test data (identical to the sparse KRR model). In short, the sparse KRR model and the SVR model are very similar. The large number of hyperparameters involved with KRR and SVR makes exact comparison impossible in practice.

An interesting experiment.

I’m old enough to have seen all the James Bond films when they were first released — even the first five with actor Sean Connery as Bond. Every one of the first five films has at least one Asian actress, even if their presence was a bit sparse. Here is a memorable scene from “Dr. No” (1962), the very first Bond film. I saw it at the Fox Fullerton (California) theater on Harbor Blvd. It is my favorite Bond film.

On the left, the woman is saying, “I am Sister Rose, this is Sister Lily.” The scene takes place shortly after Bond and Honey Ryder (actress Ursula Andress) were captured and brought to Dr. No’s underground lair. Sister Rose was played by actress Michele Mok. Sister Lily was played by actress Yvonne Shima. One of the big questions among Bond fans is what happened to Rose and Lily. It is assumed/hoped that they both escaped, as many of Dr. No’s employees did, when Bond blew up Dr. No’s facility at the end of the movie.

The sparse KRR demo program. Replace the “lt” in the accuracy() function with the less-than Boolean symbol (my blog editor chokes on symbols).

# svr_via_krr_scikit.py
# train using KRR, remove some, retrain using KRR

import numpy as np
from sklearn.kernel_ridge import KernelRidge

# KernelRidge(alpha=1, *, kernel='linear', gamma=None,
# degree=3, coef0=1, kernel_params=None)

# -----------------------------------------------------------

np.set_printoptions(precision=4, suppress=True,
  floatmode='fixed', linewidth=60)

# -----------------------------------------------------------

def accuracy(model, data_X, data_y, pct_close):
  n = len(data_X)
  n_correct = 0; n_wrong = 0
  for i in range(n):
    x = data_X[i].reshape(1,-1)
    y = data_y[i]
    y_pred = model.predict(x)[0]

    if np.abs(y - y_pred) "lt" np.abs(y * pct_close):
      n_correct += 1
    else: 
      n_wrong += 1
  return n_correct / (n_correct + n_wrong)

def mse(model, data_X, data_y):
  n = len(data_X)
  sum = 0.0
  for i in range(n):
    actual_y = data_y[i]
    pred_y = model.predict(data_X[i].reshape(1, -1))[0]
    diff = actual_y - pred_y
    sum += diff * diff
  return sum /n

# -----------------------------------------------------------
# -----------------------------------------------------------

print("\nBegin sparse KRR approximation to SVR demo ")

np.set_printoptions(precision=4, suppress=True,
    floatmode='fixed')

print("\nLoading synthetic train (200) and test (40) data ")
train_Xy = np.loadtxt(".\\Data\\synthetic_train_200.txt",
  usecols=[0,1,2,3,4,5], delimiter=",")
train_X = train_Xy[:,[0,1,2,3,4]]
train_y = train_Xy[:,5]

test_Xy = np.loadtxt(".\\Data\\synthetic_test_40.txt",
  usecols=[0,1,2,3,4,5], delimiter=",")
test_X = test_Xy[:,[0,1,2,3,4]]
test_y = test_Xy[:,5]
print("Done ")

print("\nFirst three train X: ")
for i in range(3):
  print(train_X[i])
print("\nFirst three train y: ")
for i in range(3):
  print("%0.4f " % train_y[i])

print("\nCreating preliminry KRR RBF model ")
gamma = 0.1000
alpha = 0.0010
print("Setting gamma = %0.4f " % gamma)
print("Setting alpha = %0.4f " % alpha)
model = KernelRidge(kernel='rbf', gamma=gamma, alpha=alpha)
print("Done ")

print("\nTraining KRR model ")
model.fit(train_X, train_y)
print("Done ")

print("\nRemoving non-pseudo-support vectors ")
predictions = model.predict(train_X)
residuals = np.abs(train_y - predictions)
# smaller epsilon = more support vectors
# larger epsilon = fewer support vectors
epsilon = 0.003
print("Using epsilon = %0.6f " % epsilon)
# keep only points outside epsilon-tube (support vectors)
sv_indices = np.where(residuals > epsilon)[0]
X_sv = train_X[sv_indices]
y_sv = train_y[sv_indices]
print("Number of pseudo support vectors = " + \
  str(len(X_sv)))

# retrain
print("\nRe-training sparse KRR model ")
model = KernelRidge(kernel='rbf', gamma=gamma, alpha=alpha)
model.fit(X_sv, y_sv)
print("Done ")

print("\nEvaluating model ")
acc_train = accuracy(model, X_sv, y_sv, 0.10)
acc_test = accuracy(model, test_X, test_y, 0.10)
print("\nAccuracy (within 0.10) train = %0.4f " % \
  acc_train)
print("Accuracy (within 0.10) test = %0.4f " % \
  acc_test)

mse_train = mse(model, X_sv, y_sv)
mse_test = mse(model, test_X, test_y)
print("\nMSE train = %0.4f " % mse_train)
print("MSE test = %0.4f " % mse_test)

Training data:

# synthetic_train_200.txt
#
-0.1660,  0.4406, -0.9998, -0.3953, -0.7065,  0.4840
 0.0776, -0.1616,  0.3704, -0.5911,  0.7562,  0.1568
-0.9452,  0.3409, -0.1654,  0.1174, -0.7192,  0.8054
 0.9365, -0.3732,  0.3846,  0.7528,  0.7892,  0.1345
-0.8299, -0.9219, -0.6603,  0.7563, -0.8033,  0.7955
 0.0663,  0.3838, -0.3690,  0.3730,  0.6693,  0.3206
-0.9634,  0.5003,  0.9777,  0.4963, -0.4391,  0.7377
-0.1042,  0.8172, -0.4128, -0.4244, -0.7399,  0.4801
-0.9613,  0.3577, -0.5767, -0.4689, -0.0169,  0.6861
-0.7065,  0.1786,  0.3995, -0.7953, -0.1719,  0.5569
 0.3888, -0.1716, -0.9001,  0.0718,  0.3276,  0.2500
 0.1731,  0.8068, -0.7251, -0.7214,  0.6148,  0.3297
-0.2046, -0.6693,  0.8550, -0.3045,  0.5016,  0.2129
 0.2473,  0.5019, -0.3022, -0.4601,  0.7918,  0.2613
-0.1438,  0.9297,  0.3269,  0.2434, -0.7705,  0.5171
 0.1568, -0.1837, -0.5259,  0.8068,  0.1474,  0.3307
-0.9943,  0.2343, -0.3467,  0.0541,  0.7719,  0.5581
 0.2467, -0.9684,  0.8589,  0.3818,  0.9946,  0.1092
-0.6553, -0.7257,  0.8652,  0.3936, -0.8680,  0.7018
 0.8460,  0.4230, -0.7515, -0.9602, -0.9476,  0.1996
-0.9434, -0.5076,  0.7201,  0.0777,  0.1056,  0.5664
 0.9392,  0.1221, -0.9627,  0.6013, -0.5341,  0.1533
 0.6142, -0.2243,  0.7271,  0.4942,  0.1125,  0.1661
 0.4260,  0.1194, -0.9749, -0.8561,  0.9346,  0.2230
 0.1362, -0.5934, -0.4953,  0.4877, -0.6091,  0.3810
 0.6937, -0.5203, -0.0125,  0.2399,  0.6580,  0.1460
-0.6864, -0.9628, -0.8600, -0.0273,  0.2127,  0.5387
 0.9772,  0.1595, -0.2397,  0.1019,  0.4907,  0.1611
 0.3385, -0.4702, -0.8673, -0.2598,  0.2594,  0.2270
-0.8669, -0.4794,  0.6095, -0.6131,  0.2789,  0.4700
 0.0493,  0.8496, -0.4734, -0.8681,  0.4701,  0.3516
 0.8639, -0.9721, -0.5313,  0.2336,  0.8980,  0.1412
 0.9004,  0.1133,  0.8312,  0.2831, -0.2200,  0.1782
 0.0991,  0.8524,  0.8375, -0.2102,  0.9265,  0.2150
-0.6521, -0.7473, -0.7298,  0.0113, -0.9570,  0.7422
 0.6190, -0.3105,  0.8802,  0.1640,  0.7577,  0.1056
 0.6895,  0.8108, -0.0802,  0.0927,  0.5972,  0.2214
 0.1982, -0.9689,  0.1870, -0.1326,  0.6147,  0.1310
-0.3695,  0.7858,  0.1557, -0.6320,  0.5759,  0.3773
-0.1596,  0.3581,  0.8372, -0.9992,  0.9535,  0.2071
-0.2468,  0.9476,  0.2094,  0.6577,  0.1494,  0.4132
 0.1737,  0.5000,  0.7166,  0.5102,  0.3961,  0.2611
 0.7290, -0.3546,  0.3416, -0.0983, -0.2358,  0.1332
-0.3652,  0.2438, -0.1395,  0.9476,  0.3556,  0.4170
-0.6029, -0.1466, -0.3133,  0.5953,  0.7600,  0.4334
-0.4596, -0.4953,  0.7098,  0.0554,  0.6043,  0.2775
 0.1450,  0.4663,  0.0380,  0.5418,  0.1377,  0.2931
-0.8636, -0.2442, -0.8407,  0.9656, -0.6368,  0.7429
 0.6237,  0.7499,  0.3768,  0.1390, -0.6781,  0.2185
-0.5499,  0.1850, -0.3755,  0.8326,  0.8193,  0.4399
-0.4858, -0.7782, -0.6141, -0.0008,  0.4572,  0.4197
 0.7033, -0.1683,  0.2334, -0.5327, -0.7961,  0.1776
 0.0317, -0.0457, -0.6947,  0.2436,  0.0880,  0.3345
 0.5031, -0.5559,  0.0387,  0.5706, -0.9553,  0.3107
-0.3513,  0.7458,  0.6894,  0.0769,  0.7332,  0.3170
 0.2205,  0.5992, -0.9309,  0.5405,  0.4635,  0.3532
-0.4806, -0.4859,  0.2646, -0.3094,  0.5932,  0.3202
 0.9809, -0.3995, -0.7140,  0.8026,  0.0831,  0.1600
 0.9495,  0.2732,  0.9878,  0.0921,  0.0529,  0.1289
-0.9476, -0.6792,  0.4913, -0.9392, -0.2669,  0.5966
 0.7247,  0.3854,  0.3819, -0.6227, -0.1162,  0.1550
-0.5922, -0.5045, -0.4757,  0.5003, -0.0860,  0.5863
-0.8861,  0.0170, -0.5761,  0.5972, -0.4053,  0.7301
 0.6877, -0.2380,  0.4997,  0.0223,  0.0819,  0.1404
 0.9189,  0.6079, -0.9354,  0.4188, -0.0700,  0.1907
-0.1428, -0.7820,  0.2676,  0.6059,  0.3936,  0.2790
 0.5324, -0.3151,  0.6917, -0.1425,  0.6480,  0.1071
-0.8432, -0.9633, -0.8666, -0.0828, -0.7733,  0.7784
-0.9444,  0.5097, -0.2103,  0.4939, -0.0952,  0.6787
-0.0520,  0.6063, -0.1952,  0.8094, -0.9259,  0.4836
 0.5477, -0.7487,  0.2370, -0.9793,  0.0773,  0.1241
 0.2450,  0.8116,  0.9799,  0.4222,  0.4636,  0.2355
 0.8186, -0.1983, -0.5003, -0.6531, -0.7611,  0.1511
-0.4714,  0.6382, -0.3788,  0.9648, -0.4667,  0.5950
 0.0673, -0.3711,  0.8215, -0.2669, -0.1328,  0.2677
-0.9381,  0.4338,  0.7820, -0.9454,  0.0441,  0.5518
-0.3480,  0.7190,  0.1170,  0.3805, -0.0943,  0.4724
-0.9813,  0.1535, -0.3771,  0.0345,  0.8328,  0.5438
-0.1471, -0.5052, -0.2574,  0.8637,  0.8737,  0.3042
-0.5454, -0.3712, -0.6505,  0.2142, -0.1728,  0.5783
 0.6327, -0.6297,  0.4038, -0.5193,  0.1484,  0.1153
-0.5424,  0.3282, -0.0055,  0.0380, -0.6506,  0.6613
 0.1414,  0.9935,  0.6337,  0.1887,  0.9520,  0.2540
-0.9351, -0.8128, -0.8693, -0.0965, -0.2491,  0.7353
 0.9507, -0.6640,  0.9456,  0.5349,  0.6485,  0.1059
-0.0462, -0.9737, -0.2940, -0.0159,  0.4602,  0.2606
-0.0627, -0.0852, -0.7247, -0.9782,  0.5166,  0.2977
 0.0478,  0.5098, -0.0723, -0.7504, -0.3750,  0.3335
 0.0090,  0.3477,  0.5403, -0.7393, -0.9542,  0.4415
-0.9748,  0.3449,  0.3736, -0.1015,  0.8296,  0.4358
 0.2887, -0.9895, -0.0311,  0.7186,  0.6608,  0.2057
 0.1570, -0.4518,  0.1211,  0.3435, -0.2951,  0.3244
 0.7117, -0.6099,  0.4946, -0.4208,  0.5476,  0.1096
-0.2929, -0.5726,  0.5346, -0.3827,  0.4665,  0.2465
 0.4889, -0.5572, -0.5718, -0.6021, -0.7150,  0.2163
-0.7782,  0.3491,  0.5996, -0.8389, -0.5366,  0.6516
-0.5847,  0.8347,  0.4226,  0.1078, -0.3910,  0.6134
 0.8469,  0.4121, -0.0439, -0.7476,  0.9521,  0.1571
-0.6803, -0.5948, -0.1376, -0.1916, -0.7065,  0.7156
 0.2878,  0.5086, -0.5785,  0.2019,  0.4979,  0.2980
 0.2764,  0.1943, -0.4090,  0.4632,  0.8906,  0.2960
-0.8877,  0.6705, -0.6155, -0.2098, -0.3998,  0.7107
-0.8398,  0.8093, -0.2597,  0.0614, -0.0118,  0.6502
-0.8476,  0.0158, -0.4769, -0.2859, -0.7839,  0.7715
 0.5751, -0.7868,  0.9714, -0.6457,  0.1448,  0.1175
 0.4802, -0.7001,  0.1022, -0.5668,  0.5184,  0.1090
 0.4458, -0.6469,  0.7239, -0.9604,  0.7205,  0.0779
 0.5175,  0.4339,  0.9747, -0.4438, -0.9924,  0.2879
 0.8678,  0.7158,  0.4577,  0.0334,  0.4139,  0.1678
 0.5406,  0.5012,  0.2264, -0.1963,  0.3946,  0.2088
-0.9938,  0.5498,  0.7928, -0.5214, -0.7585,  0.7687
 0.7661,  0.0863, -0.4266, -0.7233, -0.4197,  0.1466
 0.2277, -0.3517, -0.0853, -0.1118,  0.6563,  0.1767
 0.3499, -0.5570, -0.0655, -0.3705,  0.2537,  0.1632
 0.7547, -0.1046,  0.5689, -0.0861,  0.3125,  0.1257
 0.8186,  0.2110,  0.5335,  0.0094, -0.0039,  0.1391
 0.6858, -0.8644,  0.1465,  0.8855,  0.0357,  0.1845
-0.4967,  0.4015,  0.0805,  0.8977,  0.2487,  0.4663
 0.6760, -0.9841,  0.9787, -0.8446, -0.3557,  0.1509
-0.1203, -0.4885,  0.6054, -0.0443, -0.7313,  0.4854
 0.8557,  0.7919, -0.0169,  0.7134, -0.1628,  0.2002
 0.0115, -0.6209,  0.9300, -0.4116, -0.7931,  0.4052
-0.7114, -0.9718,  0.4319,  0.1290,  0.5892,  0.3661
 0.3915,  0.5557, -0.1870,  0.2955, -0.6404,  0.2954
-0.3564, -0.6548, -0.1827, -0.5172, -0.1862,  0.4622
 0.2392, -0.4959,  0.5857, -0.1341, -0.2850,  0.2470
-0.3394,  0.3947, -0.4627,  0.6166, -0.4094,  0.5325
 0.7107,  0.7768, -0.6312,  0.1707,  0.7964,  0.2757
-0.1078,  0.8437, -0.4420,  0.2177,  0.3649,  0.4028
-0.3139,  0.5595, -0.6505, -0.3161, -0.7108,  0.5546
 0.4335,  0.3986,  0.3770, -0.4932,  0.3847,  0.1810
-0.2562, -0.2894, -0.8847,  0.2633,  0.4146,  0.4036
 0.2272,  0.2966, -0.6601, -0.7011,  0.0284,  0.2778
-0.0743, -0.1421, -0.0054, -0.6770, -0.3151,  0.3597
-0.4762,  0.6891,  0.6007, -0.1467,  0.2140,  0.4266
-0.4061,  0.7193,  0.3432,  0.2669, -0.7505,  0.6147
-0.0588,  0.9731,  0.8966,  0.2902, -0.6966,  0.4955
-0.0627, -0.1439,  0.1985,  0.6999,  0.5022,  0.3077
 0.1587,  0.8494, -0.8705,  0.9827, -0.8940,  0.4263
-0.7850,  0.2473, -0.9040, -0.4308, -0.8779,  0.7199
 0.4070,  0.3369, -0.2428, -0.6236,  0.4940,  0.2215
-0.0242,  0.0513, -0.9430,  0.2885, -0.2987,  0.3947
-0.5416, -0.1322, -0.2351, -0.0604,  0.9590,  0.3683
 0.1055,  0.7783, -0.2901, -0.5090,  0.8220,  0.2984
-0.9129,  0.9015,  0.1128, -0.2473,  0.9901,  0.4776
-0.9378,  0.1424, -0.6391,  0.2619,  0.9618,  0.5368
 0.7498, -0.0963,  0.4169,  0.5549, -0.0103,  0.1614
-0.2612, -0.7156,  0.4538, -0.0460, -0.1022,  0.3717
 0.7720,  0.0552, -0.1818, -0.4622, -0.8560,  0.1685
-0.4177,  0.0070,  0.9319, -0.7812,  0.3461,  0.3052
-0.0001,  0.5542, -0.7128, -0.8336, -0.2016,  0.3803
 0.5356, -0.4194, -0.5662, -0.9666, -0.2027,  0.1776
-0.2378,  0.3187, -0.8582, -0.6948, -0.9668,  0.5474
-0.1947, -0.3579,  0.1158,  0.9869,  0.6690,  0.2992
 0.3992,  0.8365, -0.9205, -0.8593, -0.0520,  0.3154
-0.0209,  0.0793,  0.7905, -0.1067,  0.7541,  0.1864
-0.4928, -0.4524, -0.3433,  0.0951, -0.5597,  0.6261
-0.8118,  0.7404, -0.5263, -0.2280,  0.1431,  0.6349
 0.0516, -0.8480,  0.7483,  0.9023,  0.6250,  0.1959
-0.3212,  0.1093,  0.9488, -0.3766,  0.3376,  0.2735
-0.3481,  0.5490, -0.3484,  0.7797,  0.5034,  0.4379
-0.5785, -0.9170, -0.3563, -0.9258,  0.3877,  0.4121
 0.3407, -0.1391,  0.5356,  0.0720, -0.9203,  0.3458
-0.3287, -0.8954,  0.2102,  0.0241,  0.2349,  0.3247
-0.1353,  0.6954, -0.0919, -0.9692,  0.7461,  0.3338
 0.9036, -0.8982, -0.5299, -0.8733, -0.1567,  0.1187
 0.7277, -0.8368, -0.0538, -0.7489,  0.5458,  0.0830
 0.9049,  0.8878,  0.2279,  0.9470, -0.3103,  0.2194
 0.7957, -0.1308, -0.5284,  0.8817,  0.3684,  0.2172
 0.4647, -0.4931,  0.2010,  0.6292, -0.8918,  0.3371
-0.7390,  0.6849,  0.2367,  0.0626, -0.5034,  0.7039
-0.1567, -0.8711,  0.7940, -0.5932,  0.6525,  0.1710
 0.7635, -0.0265,  0.1969,  0.0545,  0.2496,  0.1445
 0.7675,  0.1354, -0.7698, -0.5460,  0.1920,  0.1728
-0.5211, -0.7372, -0.6763,  0.6897,  0.2044,  0.5217
 0.1913,  0.1980,  0.2314, -0.8816,  0.5006,  0.1998
 0.8964,  0.0694, -0.6149,  0.5059, -0.9854,  0.1825
 0.1767,  0.7104,  0.2093,  0.6452,  0.7590,  0.2832
-0.3580, -0.7541,  0.4426, -0.1193, -0.7465,  0.5657
-0.5996,  0.5766, -0.9758, -0.3933, -0.9572,  0.6800
 0.9950,  0.1641, -0.4132,  0.8579,  0.0142,  0.2003
-0.4717, -0.3894, -0.2567, -0.5111,  0.1691,  0.4266
 0.3917, -0.8561,  0.9422,  0.5061,  0.6123,  0.1212
-0.0366, -0.1087,  0.3449, -0.1025,  0.4086,  0.2475
 0.3633,  0.3943,  0.2372, -0.6980,  0.5216,  0.1925
-0.5325, -0.6466, -0.2178, -0.3589,  0.6310,  0.3568
 0.2271,  0.5200, -0.1447, -0.8011, -0.7699,  0.3128
 0.6415,  0.1993,  0.3777, -0.0178, -0.8237,  0.2181
-0.5298, -0.0768, -0.6028, -0.9490,  0.4588,  0.4356
 0.6870, -0.1431,  0.7294,  0.3141,  0.1621,  0.1632
-0.5985,  0.0591,  0.7889, -0.3900,  0.7419,  0.2945
 0.3661,  0.7984, -0.8486,  0.7572, -0.6183,  0.3449
 0.6995,  0.3342, -0.3113, -0.6972,  0.2707,  0.1712
 0.2565,  0.9126,  0.1798, -0.6043, -0.1413,  0.2893
-0.3265,  0.9839, -0.2395,  0.9854,  0.0376,  0.4770
 0.2690, -0.1722,  0.9818,  0.8599, -0.7015,  0.3954
-0.2102, -0.0768,  0.1219,  0.5607, -0.0256,  0.3949
 0.8216, -0.9555,  0.6422, -0.6231,  0.3715,  0.0801
-0.2896,  0.9484, -0.7545, -0.6249,  0.7789,  0.4370
-0.9985, -0.5448, -0.7092, -0.5931,  0.7926,  0.5402

Test data:

# synthetic_test_40.txt
#
 0.7462,  0.4006, -0.0590,  0.6543, -0.0083,  0.1935
 0.8495, -0.2260, -0.0142, -0.4911,  0.7699,  0.1078
-0.2335, -0.4049,  0.4352, -0.6183, -0.7636,  0.5088
 0.1810, -0.5142,  0.2465,  0.2767, -0.3449,  0.3136
-0.8650,  0.7611, -0.0801,  0.5277, -0.4922,  0.7140
-0.2358, -0.7466, -0.5115, -0.8413, -0.3943,  0.4533
 0.4834,  0.2300,  0.3448, -0.9832,  0.3568,  0.1360
-0.6502, -0.6300,  0.6885,  0.9652,  0.8275,  0.3046
-0.3053,  0.5604,  0.0929,  0.6329, -0.0325,  0.4756
-0.7995,  0.0740, -0.2680,  0.2086,  0.9176,  0.4565
-0.2144, -0.2141,  0.5813,  0.2902, -0.2122,  0.4119
-0.7278, -0.0987, -0.3312, -0.5641,  0.8515,  0.4438
 0.3793,  0.1976,  0.4933,  0.0839,  0.4011,  0.1905
-0.8568,  0.9573, -0.5272,  0.3212, -0.8207,  0.7415
-0.5785,  0.0056, -0.7901, -0.2223,  0.0760,  0.5551
 0.0735, -0.2188,  0.3925,  0.3570,  0.3746,  0.2191
 0.1230, -0.2838,  0.2262,  0.8715,  0.1938,  0.2878
 0.4792, -0.9248,  0.5295,  0.0366, -0.9894,  0.3149
-0.4456,  0.0697,  0.5359, -0.8938,  0.0981,  0.3879
 0.8629, -0.8505, -0.4464,  0.8385,  0.5300,  0.1769
 0.1995,  0.6659,  0.7921,  0.9454,  0.9970,  0.2330
-0.0249, -0.3066, -0.2927, -0.4923,  0.8220,  0.2437
 0.4513, -0.9481, -0.0770, -0.4374, -0.9421,  0.2879
-0.3405,  0.5931, -0.3507, -0.3842,  0.8562,  0.3987
 0.9538,  0.0471,  0.9039,  0.7760,  0.0361,  0.1706
-0.0887,  0.2104,  0.9808,  0.5478, -0.3314,  0.4128
-0.8220, -0.6302,  0.0537, -0.1658,  0.6013,  0.4306
-0.4123, -0.2880,  0.9074, -0.0461, -0.4435,  0.5144
 0.0060,  0.2867, -0.7775,  0.5161,  0.7039,  0.3599
-0.7968, -0.5484,  0.9426, -0.4308,  0.8148,  0.2979
 0.7811,  0.8450, -0.6877,  0.7594,  0.2640,  0.2362
-0.6802, -0.1113, -0.8325, -0.6694, -0.6056,  0.6544
 0.3821,  0.1476,  0.7466, -0.5107,  0.2592,  0.1648
 0.7265,  0.9683, -0.9803, -0.4943, -0.5523,  0.2454
-0.9049, -0.9797, -0.0196, -0.9090, -0.4433,  0.6447
-0.4607,  0.1811, -0.2389,  0.4050, -0.0078,  0.5229
 0.2664, -0.2932, -0.4259, -0.7336,  0.8742,  0.1834
-0.4507,  0.1029, -0.6294, -0.1158, -0.6294,  0.6081
 0.8948, -0.0124,  0.9278,  0.2899, -0.0314,  0.1534
-0.1323, -0.8813, -0.0146, -0.0697,  0.6135,  0.2386

Posted in Machine Learning, Scikit | Leave a comment