Canonical Variates for Recursive Partitioning in Data Mining
Carmela Cappelli, Claudio Conversano · COMPSTAT · 2002
This paper deals with the problem of dimension reduction in the general context of supervised statistical learning, with particular attention to data mining applications. The main goal of the proposed methodology is to improve tree based methods as prediction tool by introducing an alternative approach to data partitioning which is meant to handle large numbers of (possibly correlated) covariates. The key idea is to use suitable combinations of covariates recursively identified.