Changeset 16852 for branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis

Timestamp:

04/19/19 15:45:45 (6 years ago)

Author:

gkronber

Message:

#2847: fixed some issues that produced errors when testing

Location:

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression

Files:

• LeafModels/PreconstructedLinearModel.cs (modified) (1 diff)
• LeafTypes/LinearLeaf.cs (modified) (1 diff)
• LeafTypes/M5Leaf.cs (modified) (1 diff)
• LeafTypes/M5regLeaf.cs (modified) (1 diff)
• M5Utilities/RegressionTreeAnalyzer.cs (modified) (1 diff)
• Splitting/M5Splitter.cs (modified) (1 diff)
• Splitting/SplitterBase.cs (modified) (2 diffs)

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/LeafModels/PreconstructedLinearModel.cs

@@ -117 +117 @@
       alglib.spdmatrixcholeskysolve(aTa, n + 1, true, aTyVector, out info, out report, out coefficients);
 
-      //if cholesky calculation fails fall back to classic linear regresseion 
+      //if Cholesky calculation fails fall back to classic linear regresseion 
       if (info != 1) {
         alglib.linearmodel lm;

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/LeafTypes/LinearLeaf.cs

@@ -44 +44 @@
     #region IModelType
     public override bool ProvidesConfidence {
-      get { return false; }
+      get { return true; }
     }
     public override IRegressionModel Build(IRegressionProblemData pd, IRandom random, CancellationToken cancellationToken, out int numberOfParameters) {

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/LeafTypes/M5Leaf.cs

@@ -169 +169 @@
       }
       while (!success && tries < 100);
-      if (coefficients == null) throw new ArgumentException("No linear model could be built");
+      if (coefficients == null || coefficients.Length != n) throw new ArgumentException("No linear model could be built");
 
       intercept = yMean;

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/LeafTypes/M5regLeaf.cs

@@ -45 +45 @@
     #region IModelType
     public override bool ProvidesConfidence {
-      get { return true; }
+      get { return false; }
     }
 

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/M5Utilities/RegressionTreeAnalyzer.cs

@@ -104 +104 @@
       var training = pd.TrainingIndices.Where(x => !regressionRuleModel.Covers(pd.Dataset, x)).ToArray();
       var test = pd.TestIndices.Where(x => !regressionRuleModel.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;
+      if (training.Length > 0 || test.Length > 0) {
+        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;
+      } else notCovered = null;
 
       var training2 = pd.TrainingIndices.Where(x => regressionRuleModel.Covers(pd.Dataset, x)).ToArray();

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/Splitting/M5Splitter.cs

@@ -55 +55 @@
     protected override void AttributeSplit(IReadOnlyList<double> attValues, IReadOnlyList<double> targetValues, int minLeafSize, out int position, out double maxImpurity, out double splitValue) {
       position = 0;
-      maxImpurity = -1E20;
+      maxImpurity = double.NegativeInfinity;
       splitValue = 0.0;
       var length = targetValues.Count;

branches/2847_M5Regression/HeuristicLab.Algorithms.DataAnalysis/3.4/M5Regression/Splitting/SplitterBase.cs

@@ -33 +33 @@
   [Item("SplitterBase", "Abstract base class for splitters")]
   public abstract class SplitterBase : ParameterizedNamedItem, ISplitter {
-    public const string SplittingStateVariableName = "RuleSetState";
+    public const string SplittingStateVariableName = "SplittingState";
 
     #region Constructors & Cloning
@@ -80 +80 @@
 
     protected virtual bool DecideSplit(IRegressionProblemData splitData, int minLeafSize, out string splitAttr, out double splitValue) {
-      var bestPos = 0;
+      var bestPos = -1;
       var bestImpurity = double.MinValue;
       var bestSplitValue = 0.0;