source: branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/M5Utilities/M5Analyzer.cs @ 16847

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2017 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.Collections.Generic;
using System.Linq;
using HeuristicLab.Analysis;
using HeuristicLab.Data;
using HeuristicLab.Optimization;
using HeuristicLab.Problems.DataAnalysis;

namespace HeuristicLab.Algorithms.DataAnalysis {
  internal static class M5Analyzer {
    private const string ConditionResultName = "Condition";
    private const string CoverResultName = "Covered Instances";
    private const string CoverageDiagramResultName = "Coverage";
    private const string RuleModelResultName = "RuleModel";

    public static Dictionary<string, int> GetRuleVariableFrequences(M5RuleSetModel ruleSetModel) {
      var res = ruleSetModel.VariablesUsedForPrediction.ToDictionary(x => x, x => 0);
      foreach (var rule in ruleSetModel.Rules)
      foreach (var att in rule.SplitAttributes)
        res[att]++;
      return res;
    }

    public static Dictionary<string, int> GetTreeVariableFrequences(M5TreeModel treeModel) {
      var res = treeModel.VariablesUsedForPrediction.ToDictionary(x => x, x => 0);
      var root = treeModel.Root;
      foreach (var cur in root.EnumerateNodes().Where(x => !x.IsLeaf))
        res[cur.SplitAttribute]++;
      return res;
    }

    public static Result CreateLeafDepthHistogram(M5TreeModel treeModel) {
      var list = new List<int>();
      GetLeafDepths(treeModel.Root, 0, list);
      var row = new DataRow("Depths", "", list.Select(x => (double) x)) {
        VisualProperties = {ChartType = DataRowVisualProperties.DataRowChartType.Histogram}
      };
      var hist = new DataTable("LeafDepths");
      hist.Rows.Add(row);
      return new Result(hist.Name, hist);
    }

    public static Result CreateRulesResult(M5RuleSetModel ruleSetModel, IRegressionProblemData pd, string resultName, bool displayModels) {
      var res = new ResultCollection();
      var i = 0;
      foreach (var rule in ruleSetModel.Rules)
        res.Add(new Result("Rule" + i++, CreateRulesResult(rule, pd, displayModels, out pd)));
      return new Result(resultName, res);
    }

    public static IResult CreateCoverageDiagram(M5RuleSetModel setModel, IRegressionProblemData problemData) {
      var res = new DataTable(CoverageDiagramResultName);
      var training = CountCoverage(setModel, problemData.Dataset, problemData.TrainingIndices);
      var test = CountCoverage(setModel, problemData.Dataset, problemData.TestIndices);
      res.Rows.Add(new DataRow("Training", "", training));
      res.Rows.Add(new DataRow("Test", "", test));

      foreach (var row in res.Rows)
        row.VisualProperties.ChartType = DataRowVisualProperties.DataRowChartType.Columns;
      res.VisualProperties.XAxisMaximumFixedValue = training.Count + 1;
      res.VisualProperties.XAxisMaximumAuto = false;
      res.VisualProperties.XAxisMinimumFixedValue = 0;
      res.VisualProperties.XAxisMinimumAuto = false;
      res.VisualProperties.XAxisTitle = "Rule";
      res.VisualProperties.YAxisTitle = "Covered Instances";

      return new Result(CoverageDiagramResultName, res);
    }

    private static void GetLeafDepths(M5NodeModel n, int depth, ICollection<int> res) {
      if (n == null) return;
      if (n.Left == null && n.Right == null) res.Add(depth);
      else {
        GetLeafDepths(n.Left, depth + 1, res);
        GetLeafDepths(n.Right, depth + 1, res);
      }
    }

    private static ResultCollection CreateRulesResult(M5RuleModel m5RuleModel, IRegressionProblemData pd, bool displayModels, out IRegressionProblemData notCovered) {
      var training = pd.TrainingIndices.Where(x => !m5RuleModel.Covers(pd.Dataset, x)).ToArray();
      var test = pd.TestIndices.Where(x => !m5RuleModel.Covers(pd.Dataset, x)).ToArray();
      var data = new Dataset(pd.Dataset.DoubleVariables, pd.Dataset.DoubleVariables.Select(v => pd.Dataset.GetDoubleValues(v, training.Concat(test)).ToArray()));
      notCovered = new RegressionProblemData(data, pd.AllowedInputVariables, pd.TargetVariable);
      notCovered.TestPartition.Start = notCovered.TrainingPartition.End = training.Length;
      notCovered.TestPartition.End = training.Length + test.Length;

      var training2 = pd.TrainingIndices.Where(x => m5RuleModel.Covers(pd.Dataset, x)).ToArray();
      var test2 = pd.TestIndices.Where(x => m5RuleModel.Covers(pd.Dataset, x)).ToArray();
      var data2 = new Dataset(pd.Dataset.DoubleVariables, pd.Dataset.DoubleVariables.Select(v => pd.Dataset.GetDoubleValues(v, training2.Concat(test2)).ToArray()));
      var covered = new RegressionProblemData(data2, pd.AllowedInputVariables, pd.TargetVariable);
      covered.TestPartition.Start = covered.TrainingPartition.End = training2.Length;
      covered.TestPartition.End = training2.Length + test2.Length;

      var res = new ResultCollection {
        new Result(ConditionResultName, new StringValue(m5RuleModel.ToCompactString())),
        new Result(CoverResultName, new IntValue(pd.TrainingIndices.Count() - training.Length))
      };
      if (displayModels) res.Add(new Result(RuleModelResultName, m5RuleModel.CreateRegressionSolution(covered)));
      return res;
    }

    private static IReadOnlyList<double> CountCoverage(M5RuleSetModel setModel, IDataset data, IEnumerable<int> rows) {
      var rules = setModel.Rules.ToArray();
      var res = new double[rules.Length];
      foreach (int row in rows)
        for (int i = 0; i < rules.Length; i++)
          if (rules[i].Covers(data, row)) {
            res[i]++;
            break;
          }
      return res;
    }
  }
}