source: branches/Templates/RegressionProblemInstances/RegressionProblemInstance/RegressionProblemInstanceDescriptor.cs @ 16003

using System;
using System.Linq;
using HeuristicLab.Core;
using HeuristicLab.Problems.DataAnalysis;


namespace RegressionProblemInstances {
  // This is a general descriptor for regression problem instances which produces data for a given target expression
  public class RegressionProblemInstanceDescriptor : IRegressionProblemInstanceDescriptor {
    private IRandom rand;
    private int dim;
    private Func<double[], double> func;
    private int numRows;

    public string Name { get; }
    public string Description { get; }

    public RegressionProblemInstanceDescriptor(string name, string desc, IRandom rand, int numRows, int dim, Func<double[], double> func) {
      this.Name = name;
      this.Description = desc;
      this.rand = rand;
      this.numRows = numRows;
      this.dim = dim;
      this.func = func;
    }


    public IRegressionProblemData GenerateData() {
      var inputs = Enumerable.Range(1, dim).Select(idx => "x" + idx);
      var target = "y";

      // generate data
      var x = new double[dim]; // for evaluating the expression
      var data = new double[numRows, dim + 1];
      int i = 0;
      while (i < numRows) {
        for (int j = 0; j < dim; j++) {
          // we use the supplied PRND to generate x independently and identically distributed
          // the PRND could use any kind of probability distribution
          x[j] = rand.NextDouble();
          data[i, j] = x[j];
        }
        data[i, dim] = func(x); // y = f(x)
        // only accept reasonable values
        if (!double.IsInfinity(data[i, dim]) && !double.IsNaN(data[i, dim])) i++;
      }

      var ds = new Dataset(inputs.Concat(new string[] { target }), data);
      ds.Name = Name;
      var probData = new RegressionProblemData(ds, inputs, target);
      probData.Name = Name;
      return probData;
    }

  }
}