Changeset 712 for trunk/sources/HeuristicLab.GP.StructureIdentification.Classification

Timestamp:

11/05/08 21:34:12 (16 years ago)

Author:

gkronber

Message:

fixed a stupid mistake introduced with r702 #328 (GP evaluation doesn't work in a thread parallel engine).

Location:

trunk/sources/HeuristicLab.GP.StructureIdentification.Classification

Files:

• AccuracyEvaluator.cs (modified) (3 diffs)
• ClassificationMeanSquaredErrorEvaluator.cs (modified) (2 diffs)
• ConfusionMatrixEvaluator.cs (modified) (3 diffs)
• MulticlassModeller.cs (modified) (4 diffs)

trunk/sources/HeuristicLab.GP.StructureIdentification.Classification/AccuracyEvaluator.cs ¶

@@ -45 +45 @@
     public override void Evaluate(IScope scope, BakedTreeEvaluator evaluator, Dataset dataset, int targetVariable, double[] classes, double[] thresholds, int start, int end) {
       DoubleData accuracy = GetVariableValue<DoubleData>("Accuracy", scope, false, false);
-      if(accuracy == null) {
+      if (accuracy == null) {
         accuracy = new DoubleData();
         scope.AddVariable(new HeuristicLab.Core.Variable(scope.TranslateName("Accuracy"), accuracy));
@@ -52 +52 @@
       int nSamples = end - start;
       int nCorrect = 0;
-      for(int sample = start; sample < end; sample++) {
+      for (int sample = start; sample < end; sample++) {
         double est = evaluator.Evaluate(sample);
-        double origClass = dataset.GetValue(targetVariable, sample);
+        double origClass = dataset.GetValue(sample, targetVariable);
         double estClass = double.NaN;
         // if estimation is lower than the smallest threshold value -> estimated class is the lower class
-        if(est < thresholds[0]) estClass = classes[0];
+        if (est < thresholds[0]) estClass = classes[0];
         // if estimation is larger (or equal) than the largest threshold value -> estimated class is the upper class
-        else if(est >= thresholds[thresholds.Length - 1]) estClass = classes[classes.Length - 1];
+        else if (est >= thresholds[thresholds.Length - 1]) estClass = classes[classes.Length - 1];
         else {
           // otherwise the estimated class is the class which upper threshold is larger than the estimated value
-          for(int k = 0; k < thresholds.Length; k++) {
-            if(thresholds[k] > est) {
+          for (int k = 0; k < thresholds.Length; k++) {
+            if (thresholds[k] > est) {
               estClass = classes[k];
               break;
@@ -69 +69 @@
           }
         }
-        if(Math.Abs(estClass - origClass) < EPSILON) nCorrect++;
+        if (Math.Abs(estClass - origClass) < EPSILON) nCorrect++;
       }
       accuracy.Data = nCorrect / (double)nSamples;

trunk/sources/HeuristicLab.GP.StructureIdentification.Classification/ClassificationMeanSquaredErrorEvaluator.cs ¶

@@ -43 +43 @@
     }
 
-    public override void  Evaluate(IScope scope, BakedTreeEvaluator evaluator, HeuristicLab.DataAnalysis.Dataset dataset, int targetVariable, double[] classes, double[] thresholds, int start, int end)
-{
+    public override void Evaluate(IScope scope, BakedTreeEvaluator evaluator, HeuristicLab.DataAnalysis.Dataset dataset, int targetVariable, double[] classes, double[] thresholds, int start, int end) {
       double errorsSquaredSum = 0;
-      for(int sample = start; sample < end; sample++) {
+      for (int sample = start; sample < end; sample++) {
         double estimated = evaluator.Evaluate(sample);
-        double original = dataset.GetValue(targetVariable, sample);
-        if(!double.IsNaN(original) && !double.IsInfinity(original)) {
+        double original = dataset.GetValue(sample, targetVariable);
+        if (!double.IsNaN(original) && !double.IsInfinity(original)) {
           double error = estimated - original;
           // between classes use squared error
           // on the lower end and upper end only add linear error if the absolute error is larger than 1
           // the error>1.0 constraint is needed for balance because in the interval ]-1, 1[ the squared error is smaller than the absolute error
-          if((IsEqual(original, classes[0]) && error < -1.0) ||
+          if ((IsEqual(original, classes[0]) && error < -1.0) ||
             (IsEqual(original, classes[classes.Length - 1]) && error > 1.0)) {
             errorsSquaredSum += Math.Abs(error); // only add linear error below the smallest class or above the largest class
@@ -64 +63 @@
 
       errorsSquaredSum /= (end - start);
-      if(double.IsNaN(errorsSquaredSum) || double.IsInfinity(errorsSquaredSum)) {
+      if (double.IsNaN(errorsSquaredSum) || double.IsInfinity(errorsSquaredSum)) {
         errorsSquaredSum = double.MaxValue;
       }
 
       DoubleData mse = GetVariableValue<DoubleData>("MSE", scope, false, false);
-      if(mse == null) {
+      if (mse == null) {
         mse = new DoubleData();
         scope.AddVariable(new HeuristicLab.Core.Variable(scope.TranslateName("MSE"), mse));

trunk/sources/HeuristicLab.GP.StructureIdentification.Classification/ConfusionMatrixEvaluator.cs ¶

@@ -43 +43 @@
     public override void Evaluate(IScope scope, BakedTreeEvaluator evaluator, HeuristicLab.DataAnalysis.Dataset dataset, int targetVariable, double[] classes, double[] thresholds, int start, int end) {
       IntMatrixData matrix = GetVariableValue<IntMatrixData>("ConfusionMatrix", scope, false, false);
-      if(matrix == null) {
+      if (matrix == null) {
         matrix = new IntMatrixData(new int[classes.Length, classes.Length]);
         scope.AddVariable(new HeuristicLab.Core.Variable(scope.TranslateName("ConfusionMatrix"), matrix));
@@ -49 +49 @@
 
       int nSamples = end - start;
-      for(int sample = start; sample < end; sample++) {
+      for (int sample = start; sample < end; sample++) {
         double est = evaluator.Evaluate(sample);
-        double origClass = dataset.GetValue(targetVariable,sample);
+        double origClass = dataset.GetValue(sample, targetVariable);
         int estClassIndex = -1;
         // if estimation is lower than the smallest threshold value -> estimated class is the lower class
-        if(est < thresholds[0]) estClassIndex = 0;
+        if (est < thresholds[0]) estClassIndex = 0;
         // if estimation is larger (or equal) than the largest threshold value -> estimated class is the upper class
-        else if(est >= thresholds[thresholds.Length - 1]) estClassIndex = classes.Length - 1;
+        else if (est >= thresholds[thresholds.Length - 1]) estClassIndex = classes.Length - 1;
         else {
           // otherwise the estimated class is the class which upper threshold is larger than the estimated value
-          for(int k = 0; k < thresholds.Length; k++) {
-            if(thresholds[k] > est) {
+          for (int k = 0; k < thresholds.Length; k++) {
+            if (thresholds[k] > est) {
               estClassIndex = k;
               break;
@@ -68 +68 @@
 
         // find the first threshold index that is larger to the original value
-        int origClassIndex = classes.Length-1;
-        for(int i = 0; i < thresholds.Length; i++) {
-          if(origClass < thresholds[i]) {
+        int origClassIndex = classes.Length - 1;
+        for (int i = 0; i < thresholds.Length; i++) {
+          if (origClass < thresholds[i]) {
             origClassIndex = i;
             break;

trunk/sources/HeuristicLab.GP.StructureIdentification.Classification/MulticlassModeller.cs ¶

@@ -69 +69 @@
       binaryClassValues.Add(new DoubleData(0.0));
       binaryClassValues.Add(new DoubleData(1.0));
-      for(int i = 0; i < classValues.Count-1; i++) {
-        for(int j = i+1; j < classValues.Count; j++) {
+      for (int i = 0; i < classValues.Count - 1; i++) {
+        for (int j = i + 1; j < classValues.Count; j++) {
           Dataset dataset = new Dataset();
           dataset.Columns = origDataset.Columns;
@@ -82 +82 @@
           trainingSamplesStart = 0;
           List<double[]> rows = new List<double[]>();
-          for(int k = origTrainingSamplesStart; k < origTrainingSamplesEnd; k++) {
+          for (int k = origTrainingSamplesStart; k < origTrainingSamplesEnd; k++) {
             double[] row = new double[dataset.Columns];
             double targetValue = origDataset.GetValue(k, targetVariable);
-            if(IsEqual(targetValue, classAValue)) {
-              for(int l = 0; l < row.Length; l++) {
+            if (IsEqual(targetValue, classAValue)) {
+              for (int l = 0; l < row.Length; l++) {
                 row[l] = origDataset.GetValue(k, l);
               }
               row[targetVariable] = 0;
               rows.Add(row);
-            } else if(IsEqual(targetValue, classBValue)) {
-              for(int l = 0; l < row.Length; l++) {
+            } else if (IsEqual(targetValue, classBValue)) {
+              for (int l = 0; l < row.Length; l++) {
                 row[l] = origDataset.GetValue(k, l);
               }
@@ -101 +101 @@
           trainingSamplesEnd = rows.Count;
           validationSamplesStart = rows.Count;
-          for(int k = origValidationSamplesStart; k < origValidationSamplesEnd; k++) {
+          for (int k = origValidationSamplesStart; k < origValidationSamplesEnd; k++) {
             double[] row = new double[dataset.Columns];
             double targetValue = origDataset.GetValue(k, targetVariable);
-            if(IsEqual(targetValue, classAValue)) {
-              for(int l = 0; l < row.Length; l++) {
+            if (IsEqual(targetValue, classAValue)) {
+              for (int l = 0; l < row.Length; l++) {
                 row[l] = origDataset.GetValue(k, l);
               }
               row[targetVariable] = 0;
               rows.Add(row);
-            } else if(IsEqual(targetValue, classBValue)) {
-              for(int l = 0; l < row.Length; l++) {
+            } else if (IsEqual(targetValue, classBValue)) {
+              for (int l = 0; l < row.Length; l++) {
                 row[l] = origDataset.GetValue(k, l);
               }
@@ -122 +122 @@
           dataset.Rows = rows.Count;
           dataset.Samples = new double[dataset.Rows * dataset.Columns];
-          for(int k = 0; k < dataset.Rows; k++) {
-            for(int l = 0; l < dataset.Columns; l++) {
+          for (int k = 0; k < dataset.Rows; k++) {
+            for (int l = 0; l < dataset.Columns; l++) {
               dataset.SetValue(k, l, rows[k][l]);
             }
           }
 
-          Scope childScope = new Scope(classAValue+" vs. "+classBValue);
+          Scope childScope = new Scope(classAValue + " vs. " + classBValue);
 
           childScope.AddVariable(new HeuristicLab.Core.Variable(scope.TranslateName(TARGETCLASSVALUES), binaryClassValues));