Feature Selection in Marketing Applications

The paper is concerned with marketing applications of classification analysis. Feature selection (PS) is crucial in this domain to avoid cognitive overload of decision makers through use of excessively large attribute sets. Whereas algorithms for feature ranking have received considerable attention within the literature, a clear strategy how a subset of attributes should be selected once a ranking has been obtained is yet missing. Consequently; three candidate FS procedures are presented and contrasted by means of empirical experimentation on real-world data. The results offer some guidance which approach should be employed in practical applications and identify promising avenues for future research.