Changeset 4125 for trunk

Timestamp:

08/01/10 18:08:04 (14 years ago)

Author:

gkronber

Message:

Made variable frequency analyzer more efficient and removed subtraction of base line (variable impacts are now in the range 0..1 instead of -1 .. 1) #1011

Location:

trunk/sources

Files:

• HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/BestSymbolicRegressionSolutionAnalyzer.cs (modified) (1 diff)
• HeuristicLab.Problems.DataAnalysis/3.3/Symbolic/VariableFrequencyAnalyser.cs (modified) (3 diffs)

trunk/sources/HeuristicLab.Problems.DataAnalysis.Regression/3.3/Symbolic/Analyzers/BestSymbolicRegressionSolutionAnalyzer.cs

@@ -124 +124 @@
         foreach (var dataRow in variableFrequencies.Rows) {
           string variableName = dataRow.Name;
-          double integral = 0;
-          if (dataRow.Values.Count > 1) {
-            double baseline = dataRow.Values.First();
-            integral = (from value in dataRow.Values
-                        select value - baseline)
-                              .Sum();
-            integral /= dataRow.Values.Count;
-          }
-          impacts[rowIndex++, 0] = integral;
+          impacts[rowIndex++, 0] = dataRow.Values.Average();
         }
         return impacts;

trunk/sources/HeuristicLab.Problems.DataAnalysis/3.3/Symbolic/VariableFrequencyAnalyser.cs

@@ -84 +84 @@
 
     public static IEnumerable<KeyValuePair<string, double>> CalculateVariableFrequencies(IEnumerable<SymbolicExpressionTree> trees, IEnumerable<string> inputVariables) {
-      int totalVariableReferences = 0;
       Dictionary<string, double> variableReferencesSum = new Dictionary<string, double>();
       Dictionary<string, double> variableFrequencies = new Dictionary<string, double>();
@@ -94 +93 @@
           variableReferencesSum[pair.Key] += pair.Value;
         }
-        totalVariableReferences += GetTotalVariableReferencesCount(tree);
       }
+      double totalVariableReferences = variableReferencesSum.Values.Sum();
       foreach (string inputVariable in inputVariables) {
-        double relFreq = variableReferencesSum[inputVariable] / (double)totalVariableReferences;
+        double relFreq = variableReferencesSum[inputVariable] / totalVariableReferences;
         variableFrequencies.Add(inputVariable, relFreq);
       }
@@ -103 +102 @@
     }
 
-    private static int GetTotalVariableReferencesCount(SymbolicExpressionTree tree) {
-      return tree.IterateNodesPrefix().OfType<VariableTreeNode>().Count();
-    }
-
     private static IEnumerable<KeyValuePair<string, int>> GetVariableReferenceCount(SymbolicExpressionTree tree, IEnumerable<string> inputVariables) {
       Dictionary<string, int> references = new Dictionary<string, int>();
-      var groupedFuns = (from node in tree.IterateNodesPrefix().OfType<VariableTreeNode>()
-                         select node.VariableName)
-                         .GroupBy(x => x)
-                         .Select(g => new { Key = g.Key, Count = g.Count() })
-                         .ToArray();
+      var variableNames = from node in tree.IterateNodesPrefix().OfType<VariableTreeNode>()
+                          select node.VariableName;
 
-      foreach (var inputVariable in inputVariables) {
-        var matchingFuns = from g in groupedFuns
-                           where g.Key == inputVariable
-                           select g.Count;
-        if (matchingFuns.Count() == 0)
-          references.Add(inputVariable, 0);
-        else {
-          references.Add(inputVariable, matchingFuns.Single());
+      foreach (var variableName in variableNames) {
+        if (!references.ContainsKey(variableName)) {
+          references[variableName] = 1;
+        } else {
+          references[variableName] += 1;
         }
       }