Optimal classifier selection and negative bias in error rate estimation: an empirical study on high-dimensional prediction

… or in prediction modelling/supervised learning, see a paper I wrote in my young years (2009) with C. Strobl: https://t.co/00Yndztff2 4/6

@MaartenvSmeden Are you referring to the fact that CV says nothing about external validation? Or to cherry-picking of the best CV-AUC estimate out of many models (which amounts to a form of overfitting) as discussed here: https://t.co/00Yndztff2?

RT @BoulesteixLaure: @MaartenvSmeden Yes, see for example: https://t.co/00Yndztff2

@GordonJForbes @MaartenvSmeden Yes, or an independent test data set. See: https://t.co/00Yndztff2