TITLE: Correlation Pursuit: forward stepwise variable selection for index models

SPEAKER:  Michael Zhu

ABSTRACT:

In this talk, a stepwise procedure, correlation pursuit (COP), is proposed for variable selection under the sufficient dimension reduction framework. Unlike linear stepwise regression, COP does not impose assumptions on the exact form of the relationship between the response variable and the predictor variables. The COP procedure selects variables that attain the maximum correlation between the transformed response and the linear combination of the variables. Some asymptotic properties of the COP procedure are established, and in particular, its variable selection performance under diverging sample size and number of predictors has been investigated. The empirical performance of the COP procedure in comparison with existing methods are demonstrated by both simulation studies and an real life example in functional genomics.