Identification of Changes in Mean with Regression Trees: An Application to Market Research

Abstract

In this paper we present a computationally e cient method for nding multiple structural breaks at unknown dates based on regression trees. We outline the procedure and present the results of a simulation study to assess the performance of the method and to compare it with the procedure proposed by Bai and Perron. We nd the tree-based method performs well in long series which are impractical to analyze with current methods. We apply these methods plus the CUSUM test to the market share of Crest toothpaste between 1958 and 1963.