Changeset 4109 for trunk/sources/HeuristicLab.Problems.DataAnalysis.FeatureSelection/3.3/Evaluators

Timestamp:

07/27/10 09:40:14 (14 years ago)

Author:

gkronber

Message:

Use MSE instead of RMSE in feature selection for linear regression to make the results comparable with all other data analysis results. #1097

File:

• trunk/sources/HeuristicLab.Problems.DataAnalysis.FeatureSelection/3.3/Evaluators/LinearRegressionFeatureSelectionEvaluator.cs (modified) (1 diff)

trunk/sources/HeuristicLab.Problems.DataAnalysis.FeatureSelection/3.3/Evaluators/LinearRegressionFeatureSelectionEvaluator.cs

@@ -84 +84 @@
         var tree = LinearRegressionSolutionCreator.CreateSymbolicExpressionTree(dataset, targetVariable, allowedInputVariables, start, end, out rmsError, out cvRmsError);
         featureCount = allowedInputVariables.Count;
-        mse = rmsError;
+        mse = cvRmsError * cvRmsError;
       } else {
         featureCount = 0;