Employing Linear Regression in Regression Tree Leaves 1992

Abstract:  . The advantage of using linear regression in the leaves of a regression tree is analysed in the paper. It is carried out how this modification affects the construction, pruning and interpretation of a regression tree. The modification is tested on artificial and real-life domains. The results show that the modification is beneficial as it leads to smaller classification errors of induced regression trees. Keywords: machine learning, TDIDT, regression, linear regression, Bayesian approach. 1 Introduction Regression trees, similar to classification trees, are used when learning a relation between attributes and a continuous class. Their inner nodes are labelled with a test on the value of an attribute, and their leaves are labelled with a function prescribing a value to the class. A regression tree actually implements a function y(x1 ; x2 ; : : : ; xn) of n continuous or discrete attributes. In the basic CART algorithm for the construction of regression trees [1], the class value in t...