Controlling the False Discovery Rate via knockoffs: is the +1 needed?

04/28/2022

∙

by   Andrew Rajchert, et al.

∙

0

∙

share

Barber and Candès (2015) control of the FDR in feature selection relies on estimating the FDR by the number of knockoff wins +1 divided by the number of original wins. We show that the +1 is necessary in typical scenarios.