Speaker: Yushan Gu, PhD Candidate, Department of Statistics, Iowa State University

Title: F statistics for high-dimensional inference of linear model (FHDI) and its applications

Abstract: We proposed a novel method -- the F statistics for high-dimensional inference of linear model (FHDI), aims to estimate and conduct inference on the effects of multiple covariates of interest simultaneously in both single-response or multi-responses model, adjusting for the effects of high-dimensional control variables under a linear model. This is an F-type statistic based on the residuals from fitting the response variable and the target covariates to the control covariates via regularized estimation. Procedures for hypothesis testing and confidence region construction are developed for single response model.

Real data analyses of S&P 500 stock returns are conducted to demonstrate the utility of the proposed method in practice.