Research Of False Discovery Rate Control Based On Improved SDA Method

Multiple hypothesis testing is a basic problem in high-dimensional statistical inference,which is widely used in many scientific fields.In this paper,we discuss how to realize accurate FDR control in multiple testing under the condition of correlation.SDA method proposed by Du and Guo is discussed^[51],which is a distribution–free FDR control method.Under the condition of correlation,it has obvious advantages in FDR control.However,it is affected by data sparsity level,signal amplitude and feature correlation structures,the FDR control of SDA method becomes less accurate due to the additional estimation errors.In this paper,SDA method is improved based on the SLOPE variable selection method and MEWMA control chart respectively,and the differences between FDP estimated values obtained by these three methods and the corresponding FDP true values are analyzed visually.the effectiveness and robustness of the three methods are verified by simulation experiment.