New quality measures for ClassificationSolutions

[mkommend]

The sensitivity and specificity should be calculated for classification solutions. Further interesting measures would be the precision, negative predictive value, false positive rate, or the false discovery rate.

​http://en.wikipedia.org/wiki/Sensitivity_and_specificity

[mkommend]

r11603: Branched HL.Problems.DataAnalysis to implement the new quality measures for classification solutions.

[mkommend]

r11604: Adapted solution and project settings in ClassificationExtensions branch.

[mkommend]

r11606: Added the possibility to specify a positive class in ClassificationProblemData.

[mkommend]

r11607: Adapted IClassificationProblemData interface (added positive class name).

[ehopf]

r11622: Implemented classification quality measures

• true positive rate
• true negative rate
• positive predictive value
• negative predictive value
• false positive rate
• false discovery rate

[mkommend]

r11643: Merged trunk changes (update to .NET 4.5)

[ehopf]

r11683:

• Classification performance measures improvements in terms of readablity of the code.
• Added positive class value to the classification performance measures result collection.
• Fixed bug: classification performance measures will now be calculated after loading from files with non serialized performance measures results.

[ehopf]

r11684: Encapsulated the classification performance measures to the ClassificationPerformanceMeasuresResultCollection

[ehopf]

r11685: Improvements on the classification performance measures encapsulation

[ehopf]

r11686: Added missing Exception descriptions.

[ehopf]

r11687: Branched HeuristicLab.Optimization

[ehopf]

r11688: Adapted project and solution settings in ClassificationExtensions branch.

[ehopf]

r11689: Added methods to collect result values to the ResultCollection-class.

[ehopf]

r11718: Replaced the collect result values method in the ResultCollection-class.

[ehopf]

r11719: Changed the used collect results values method to the new method in the Algorithm-class.

[ehopf]

r11720: Fixed infinite loop.

[mkommend]

r11761: Minor code cleanup in classification extensions.

[mkommend]

r11762: Merged changes in HL.Optimization into the trunk. r11763: Merged changes in HL.Problems.DataAnalysis into the trunk.

For a stable release the two changesets above must be merged into stable.

[mkommend]

r11764: Deleted branch for classification extensions.

[mkommend]

r11766: Renamed positive class parameter to satisfy the parameter visibility unit test and fixed cloning of classification performance measures.

[mkommend]

r11872: Merged r11762, r11763, r11764 and r11766 into stable.