source: branches/PersistentDataStructures/HeuristicLab.ExtLibs/HeuristicLab.LibSVM/3.12/LibSVM-3.12/RangeTransform.cs @ 14710

/*
 * SVM.NET Library
 * Copyright (C) 2008 Matthew Johnson
 * 
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 * 
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 * 
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 * 
 * Adaptions to work with direct C# translation of the java libSVM source code (version 3.1.2) by Gabriel Kronberger
 * 
 */


using System;
using System.Globalization;
using System.IO;
using System.Linq;
using System.Threading;

namespace LibSVM {
  /// <summary>
  /// Class which encapsulates a range transformation.
  /// </summary>
  public class RangeTransform {
    /// <summary>
    /// Default lower bound for scaling (-1).
    /// </summary>
    public const int DEFAULT_LOWER_BOUND = -1;
    /// <summary>
    /// Default upper bound for scaling (1).
    /// </summary>
    public const int DEFAULT_UPPER_BOUND = 1;

    /// <summary>
    /// Determines the Range transform for the provided problem.  Uses the default lower and upper bounds.
    /// </summary>
    /// <param name="prob">The Problem to analyze</param>
    /// <returns>The Range transform for the problem</returns>
    public static RangeTransform Compute(svm_problem prob) {
      return Compute(prob, DEFAULT_LOWER_BOUND, DEFAULT_UPPER_BOUND);
    }
    /// <summary>
    /// Determines the Range transform for the provided problem.
    /// </summary>
    /// <param name="prob">The Problem to analyze</param>
    /// <param name="lowerBound">The lower bound for scaling</param>
    /// <param name="upperBound">The upper bound for scaling</param>
    /// <returns>The Range transform for the problem</returns>
    public static RangeTransform Compute(svm_problem prob, double lowerBound, double upperBound) {
      // node indices must be ordered
      int maxIndex = prob.x.Select(arr => arr.Last().index).Max();

      double[] minVals = new double[maxIndex];
      double[] maxVals = new double[maxIndex];
      int[] count = new int[maxIndex];
      for (int i = 0; i < maxIndex; i++) {
        minVals[i] = double.MaxValue;
        maxVals[i] = double.MinValue;
        count[i] = 0;
      }
      for (int i = 0; i < prob.l; i++) {
        for (int j = 0; j < prob.x[i].Length; j++) {
          int index = prob.x[i][j].index - 1;
          double value = prob.x[i][j].value;
          minVals[index] = Math.Min(minVals[index], value);
          maxVals[index] = Math.Max(maxVals[index], value);
          count[index]++;
        }
      }
      for (int i = 0; i < maxIndex; i++) {
        if (count[i] == 0) {
          minVals[i] = 0;
          maxVals[i] = 0;
        }
      }
      return new RangeTransform(minVals, maxVals, lowerBound, upperBound);
    }

    private double[] _inputStart;
    private double[] _inputScale;
    private double _outputStart;
    private double _outputScale;
    private int _length;

    /// <summary>
    /// Constructor.
    /// </summary>
    /// <param name="minValues">The minimum values in each dimension.</param>
    /// <param name="maxValues">The maximum values in each dimension.</param>
    /// <param name="lowerBound">The desired lower bound for all dimensions.</param>
    /// <param name="upperBound">The desired upper bound for all dimensions.</param>
    public RangeTransform(double[] minValues, double[] maxValues, double lowerBound, double upperBound) {
      _length = minValues.Length;
      if (maxValues.Length != _length)
        throw new Exception("Number of max and min values must be equal.");
      _inputStart = new double[_length];
      _inputScale = new double[_length];
      for (int i = 0; i < _length; i++) {
        _inputStart[i] = minValues[i];
        _inputScale[i] = maxValues[i] - minValues[i];
      }
      _outputStart = lowerBound;
      _outputScale = upperBound - lowerBound;
    }
    private RangeTransform(double[] inputStart, double[] inputScale, double outputStart, double outputScale, int length) {
      _inputStart = inputStart;
      _inputScale = inputScale;
      _outputStart = outputStart;
      _outputScale = outputScale;
      _length = length;
    }
    /// <summary>
    /// Transforms the input array based upon the values provided.
    /// </summary>
    /// <param name="input">The input array</param>
    /// <returns>A scaled array</returns>
    public svm_node[] Transform(svm_node[] input) {
      svm_node[] output = new svm_node[input.Length];
      for (int i = 0; i < output.Length; i++) {
        int index = input[i].index;
        double value = input[i].value;
        output[i] = new svm_node() { index = index, value = Transform(value, index) };
      }
      return output;
    }

    /// <summary>
    /// Transforms this an input value using the scaling transform for the provided dimension.
    /// </summary>
    /// <param name="input">The input value to transform</param>
    /// <param name="index">The dimension whose scaling transform should be used</param>
    /// <returns>The scaled value</returns>
    public double Transform(double input, int index) {
      index--;
      double tmp = input - _inputStart[index];
      if (Math.Abs(_inputScale[index]) < 1E-12)
        return 0;
      tmp /= _inputScale[index];
      tmp *= _outputScale;
      return tmp + _outputStart;
    }
    /// <summary>
    /// Writes this Range transform to a stream.
    /// </summary>
    /// <param name="stream">The stream to write to</param>
    /// <param name="r">The range to write</param>
    public static void Write(Stream stream, RangeTransform r) {
      var savedCulture = Thread.CurrentThread.CurrentCulture;
      Thread.CurrentThread.CurrentCulture = CultureInfo.InvariantCulture;


      StreamWriter output = new StreamWriter(stream);
      output.WriteLine(r._length);
      output.Write(r._inputStart[0].ToString("r"));
      for (int i = 1; i < r._inputStart.Length; i++)
        output.Write(" " + r._inputStart[i].ToString("r"));
      output.WriteLine();
      output.Write(r._inputScale[0].ToString("r"));
      for (int i = 1; i < r._inputScale.Length; i++)
        output.Write(" " + r._inputScale[i].ToString("r"));
      output.WriteLine();
      output.WriteLine("{0} {1}", r._outputStart.ToString("r"), r._outputScale.ToString("r"));
      output.Flush();

      Thread.CurrentThread.CurrentCulture = savedCulture;
    }

    /// <summary>
    /// Writes this Range transform to a file.    This will overwrite any previous data in the file.
    /// </summary>
    /// <param name="outputFile">The file to write to</param>
    /// <param name="r">The Range to write</param>
    public static void Write(string outputFile, RangeTransform r) {
      FileStream s = File.Open(outputFile, FileMode.Create);
      try {
        Write(s, r);
      }
      finally {
        s.Close();
      }
    }

    /// <summary>
    /// Reads a Range transform from a file.
    /// </summary>
    /// <param name="inputFile">The file to read from</param>
    /// <returns>The Range transform</returns>
    public static RangeTransform Read(string inputFile) {
      FileStream s = File.OpenRead(inputFile);
      try {
        return Read(s);
      }
      finally {
        s.Close();
      }
    }

    /// <summary>
    /// Reads a Range transform from a stream.
    /// </summary>
    /// <param name="stream">The stream to read from</param>
    /// <returns>The Range transform</returns>
    public static RangeTransform Read(Stream stream) {
      return Read(new StreamReader(stream));
    }

    public static RangeTransform Read(TextReader input) {
      var savedCulture = Thread.CurrentThread.CurrentCulture;
      Thread.CurrentThread.CurrentCulture = CultureInfo.InvariantCulture;

      int length = int.Parse(input.ReadLine());
      double[] inputStart = new double[length];
      double[] inputScale = new double[length];
      string[] parts = input.ReadLine().Split();
      for (int i = 0; i < length; i++)
        inputStart[i] = double.Parse(parts[i]);
      parts = input.ReadLine().Split();
      for (int i = 0; i < length; i++)
        inputScale[i] = double.Parse(parts[i]);
      parts = input.ReadLine().Split();
      double outputStart = double.Parse(parts[0]);
      double outputScale = double.Parse(parts[1]);

      Thread.CurrentThread.CurrentCulture = savedCulture;

      return new RangeTransform(inputStart, inputScale, outputStart, outputScale, length);
    }
  }
}