Overfitting in Making Comparisons Between Variable Selection MethodsOverfitting in Making Comparisons Between Variable Selection MethodsReunanen, Juha2003
Paper summaryjoecohenThis paper discusses an important bias in evaluation of methods using cross-validation. A method that makes decisions based of cross validation can appear to increase overall performance by simply dealing with the bias of cross-validation and not the real problem.
This paper discusses an important bias in evaluation of methods using cross-validation. A method that makes decisions based of cross validation can appear to increase overall performance by simply dealing with the bias of cross-validation and not the real problem.