Impala Documentation

virtual void Impala::Core::Training::TrainDataSrc::FilterTestFold	(	int	f,
		int	foldCount,
		int	repetition,
		bool	episodeConstrained
	)			`[inline, virtual]`

Definition at line 85 of file TrainDataSrc.h.

References MakeFolds(), mSelection, and Impala::Core::Table::Sort().

Referenced by Impala::Core::Training::ClassifierEvaluator::Evaluate(), and Impala::Core::Training::Tester::FirstFold().

00086     {
00087         std::vector<Table::QuidTable*> folds =
00088             MakeFolds(f, foldCount, repetition, episodeConstrained);
00089         for(int i=0 ; i<folds.size() ; ++i)
00090         {
00091             if(f == i)
00092                 mSelection = folds[f];
00093             else
00094                 delete folds[i];
00095         }
00096         // I'm actually not sure whether a sort here makes sense:
00097         // - on the one hand it makes TrainDataSrc more uniform and easier to debug
00098         // - on the other hand it is unnessecary and a bit slower
00099         // Let's stick with debuggability for now 
00100         Sort(mSelection, 1, true);
00101     }

Here is the call graph for this function: