Abstract:

Feature screening in ultrahigh-dimensional varying coefficient models is a crucial statistical problem in economics, genomics and etc. Existing methods suffer in the cases of multiple index variables and group predictor variables. Moreover, current methods can not handle nonlinear varying coefficient models which is possible in reality. To deal with those scenarios efficiently in real life, we develop a screening procedure for ultrahigh-dimensional varying coefficient models utilizing conditional distance covariance (CDC). Extensive simulation studies and two real economic data examples have shown the effectiveness and the flexibility of our proposed methods.