Estimating central subspaces via inverse third moments

Modern graphical tools have enhanced our ability to learn many things from data directly. In recent years, dimension reduction has proven to be an effective tool for generating low-dimensional summary plots without appreciable loss of information. Some well-known inverse regression methods for dimension reduction such as sliced inverse regression (Li, 1991) and sliced average variance estimation (Cook & Weisberg, 1991) have been developed to estimate summary plots for regression and discriminant analysis. In this paper, we suggest a new method that makes use of inverse third moments. This method can find structure beyond that found by sliced inverse regression and sliced average variance estimation, particularly regression mixtures. Illustrative examples are presented.