source: branches/RegressionBenchmarks/HeuristicLab.Problems.DataAnalysis.Views/3.4/Regression/RegressionSolutionResidualHistogram.cs @ 7255

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2011 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab 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.
 *
 * HeuristicLab 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 HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System;
using System.Collections.Generic;
using System.Drawing;
using System.Linq;
using System.Windows.Forms;
using System.Windows.Forms.DataVisualization.Charting;
using HeuristicLab.MainForm;
using HeuristicLab.MainForm.WindowsForms;

namespace HeuristicLab.Problems.DataAnalysis.Views {
  [View("Residual Histogram")]
  [Content(typeof(IRegressionSolution))]
  public partial class RegressionSolutionResidualHistogram : DataAnalysisSolutionEvaluationView {
    protected const string ALL_SAMPLES = "All samples";
    protected const string TRAINING_SAMPLES = "Training samples";
    protected const string TEST_SAMPLES = "Test samples";

    //to reduce code duplication
    protected static string[] ALL_SERIES = new string[] { ALL_SAMPLES, TRAINING_SAMPLES, TEST_SAMPLES };

    //approximate amount of bins
    protected const double bins = 25;

    /*
     * keeps for all series a list for all bins with the position of the bin, the relative frequency of the
     * residuals and the beginning and the end of the interval of the bin
     */
    protected Dictionary<string, List<List<double>>> relativeFrequencies;

    public new IRegressionSolution Content {
      get { return (IRegressionSolution)base.Content; }
      set { base.Content = value; }
    }

    public RegressionSolutionResidualHistogram()
      : base() {
      InitializeComponent();

      relativeFrequencies = new Dictionary<string, List<List<double>>>();

      foreach (string series in ALL_SERIES) {
        this.chart.Series.Add(series);
        this.chart.Series[series].LegendText = series;
        this.chart.Series[series].ChartType = SeriesChartType.Column;
        this.chart.Series[series]["PointWidth"] = "0.9";
        this.chart.Series[series].BorderWidth = 1;
        this.chart.Series[series].BorderDashStyle = ChartDashStyle.Solid;
        this.chart.Series[series].BorderColor = Color.Black;
        this.chart.Series[series].ToolTip = series + " Y = #VALY from #CUSTOMPROPERTY(from) to #CUSTOMPROPERTY(to)";

        relativeFrequencies[series] = new List<List<double>>();
      }

      //configure axis 
      this.chart.CustomizeAllChartAreas();
      this.chart.ChartAreas[0].AxisX.Title = "Residuals";
      this.chart.ChartAreas[0].CursorX.IsUserSelectionEnabled = true;
      this.chart.ChartAreas[0].AxisX.ScaleView.Zoomable = true;
      this.chart.ChartAreas[0].CursorX.Interval = 1;
      this.chart.ChartAreas[0].CursorY.Interval = 1;

      this.chart.ChartAreas[0].AxisY.Title = "Relative Frequency";
      this.chart.ChartAreas[0].CursorY.IsUserSelectionEnabled = true;
      this.chart.ChartAreas[0].AxisY.ScaleView.Zoomable = true;
      this.chart.ChartAreas[0].AxisY.IsStartedFromZero = true;
    }

    private void RedrawChart() {

      foreach (string series in ALL_SERIES) {
        this.chart.Series[series].Points.Clear();
        this.relativeFrequencies[series].Clear();
      }

      if (Content != null) {

        Dictionary<string, List<double>> residuals = CalculateResiduals();

        double max = Math.Max(Math.Abs(residuals[ALL_SAMPLES].Min()), Math.Abs(residuals[ALL_SAMPLES].Max()));
        double intervalWidth = (max * 2.0) / bins;

        max = HumanRoundMax(max);
        intervalWidth = HumanRoundMax(intervalWidth);

        foreach (string series in ALL_SERIES) {
          CalculateFrequencies(residuals[series], series, max, intervalWidth);
        }

        ChartArea chartArea = chart.ChartAreas[0];

        chartArea.AxisX.Minimum = -max - intervalWidth;
        chartArea.AxisX.Maximum = max + intervalWidth;
        chartArea.AxisY.Maximum = relativeFrequencies[ALL_SAMPLES].Select(x => x.ElementAt(1)).Max() + 0.02;
        chartArea.AxisY.Interval = 0.1;

        chartArea.AxisX.Interval = intervalWidth;
        int curBins = (int)Math.Round(max / intervalWidth);
        //shifts the x axis label so that zero is in the middle
        if (curBins % 2 == 0)
          chartArea.AxisX.IntervalOffset = intervalWidth;
        else
          chartArea.AxisX.IntervalOffset = intervalWidth / 2;
      }
    }

    private Dictionary<string, List<double>> CalculateResiduals() {
      Dictionary<string, List<double>> residuals = new Dictionary<string, List<double>>();

      foreach (string series in ALL_SERIES) {
        residuals[series] = new List<double>();
      }

      IRegressionProblemData problemdata = Content.ProblemData;
      List<double> targetValues = problemdata.Dataset.GetDoubleValues(Content.ProblemData.TargetVariable).ToList();
      List<double> estimatedValues = Content.EstimatedValues.ToList();

      for (int i = 0; i < Content.ProblemData.Dataset.Rows; i++) {
        double residual = estimatedValues[i] - targetValues[i];
        residuals[ALL_SAMPLES].Add(residual);
        if (i >= problemdata.TrainingPartition.Start && i < problemdata.TrainingPartition.End)
          residuals[TRAINING_SAMPLES].Add(residual);
        if (i >= problemdata.TestPartition.Start && i < problemdata.TestPartition.End)
          residuals[TEST_SAMPLES].Add(residual);
      }

      return residuals;
    }

    private void CalculateFrequencies(List<double> residualValues, string series, double max, double intervalWidth) {
      Series residualSeries = this.chart.Series[series];

      double intervalCenter = intervalWidth / 2.0;
      double sampleCount = residualValues.Count();
      double current = -max;

      for (int i = 0; i <= bins; i++) {
        IEnumerable<double> help = residualValues.Where(x => x >= (current - intervalCenter) && x < (current + intervalCenter));
        relativeFrequencies[series].Add(new List<double>() { current, help.Count() / sampleCount, current - intervalCenter, current + intervalCenter });
        current += intervalWidth;
      }

      ShowValues(residualSeries, relativeFrequencies[series]);
    }

    private void ToggleSeriesData(Series series) {
      if (series.Points.Count > 0) {  //checks if series is shown
        if (this.chart.Series.Any(s => s != series && s.Points.Count > 0)) {
          series.Points.Clear();
        }
      } else if (Content != null) {
        ShowValues(series, relativeFrequencies[series.Name]);
        chart.Legends[series.Legend].ForeColor = Color.Black;
        chart.Refresh();
      }
    }

    private void ShowValues(Series series, List<List<double>> relativeSeriesFrequencies) {
      DataPointCollection seriesPoints = series.Points;

      foreach (var valueList in relativeSeriesFrequencies) {
        seriesPoints.AddXY(valueList[0], valueList[1]);
        seriesPoints[seriesPoints.Count - 1]["from"] = valueList[2].ToString();
        seriesPoints[seriesPoints.Count - 1]["to"] = valueList[3].ToString();
      }
    }

    private double HumanRoundMax(double max) {
      double base10;
      if (max > 0) base10 = Math.Pow(10.0, Math.Floor(Math.Log10(max)));
      else base10 = Math.Pow(10.0, Math.Ceiling(Math.Log10(-max)));
      double rounding = (max > 0) ? base10 : -base10;
      while (rounding < max) rounding += base10;
      return rounding;
    }

    #region events
    protected override void RegisterContentEvents() {
      base.RegisterContentEvents();
      Content.ModelChanged += new EventHandler(Content_ModelChanged);
      Content.ProblemDataChanged += new EventHandler(Content_ProblemDataChanged);
    }
    protected override void DeregisterContentEvents() {
      base.DeregisterContentEvents();
      Content.ModelChanged -= new EventHandler(Content_ModelChanged);
      Content.ProblemDataChanged -= new EventHandler(Content_ProblemDataChanged);
    }

    protected override void OnContentChanged() {
      base.OnContentChanged();
      RedrawChart();
    }
    private void Content_ProblemDataChanged(object sender, EventArgs e) {
      RedrawChart();
    }
    private void Content_ModelChanged(object sender, EventArgs e) {
      RedrawChart();
    }

    private void chart_MouseDown(object sender, MouseEventArgs e) {
      HitTestResult result = chart.HitTest(e.X, e.Y);
      if (result.ChartElementType == ChartElementType.LegendItem) {
        ToggleSeriesData(result.Series);
      }
    }
    private void chart_MouseMove(object sender, MouseEventArgs e) {
      HitTestResult result = chart.HitTest(e.X, e.Y);
      if (result.ChartElementType == ChartElementType.LegendItem)
        Cursor = Cursors.Hand;
      else
        Cursor = Cursors.Default;
    }
    private void chart_CustomizeLegend(object sender, CustomizeLegendEventArgs e) {
      if (chart.Series.Count != 3) return;
      e.LegendItems[0].Cells[1].ForeColor = this.chart.Series[ALL_SAMPLES].Points.Count == 0 ? Color.Gray : Color.Black;
      e.LegendItems[1].Cells[1].ForeColor = this.chart.Series[TRAINING_SAMPLES].Points.Count == 0 ? Color.Gray : Color.Black;
      e.LegendItems[2].Cells[1].ForeColor = this.chart.Series[TEST_SAMPLES].Points.Count == 0 ? Color.Gray : Color.Black;
    }
    #endregion
  }
}