An Approach to Learn Structural Similarity Between Decision Trees Using Hungarian Algorithm

Abstract

The prevailing data mining techniques emphasize on discovering patterns and extract useful information from facts already recorded in various databases. Prior research substantiates the belief that the decision tree methodology is one of the most prevalent technique for classification systems, predictive modeling, and interpretation as well as data manipulations. In this paper, we intend to delve deeper to find out structural similarity measure among decision trees induced from dataset using a naive approach. The purpose of this approach is to quantify the similarity which is an important factor, especially for data clustering. Decision tree similarity can be broadly classified under two categories (1) Structural Similarity (2) Semantic Similarity. Thus, the main focus of this paper is to determine the homogeneity between decision trees as the similarity measures are the key contributors in solving pattern recognition problems.