Weighted Pattern Trees: A Case Study with Customer Satisfaction Dataset

Abstract

A pattern tree [1] is a tree which propagates fuzzy terms using different fuzzy aggregations. Each pattern tree represents a structure for an output class in the sense that how the fuzzy terms aggregate to predict such a class. Unlike decision trees, pattern trees explicitly make use of t-norms (i.e., AND) and t-conorms (OR) to build trees, which is essential for applications requiring rules connected with t-conorms explicitly. Pattern trees can not only obtain high accuracy rates in classification applications, but also be robust to over-fitting. This paper further extends pattern trees approach by assigning certain weights to different trees, to reflect the nature that different trees may have different confidences. The concept of weighted pattern trees is important as it offers an option to trade off the complexity and performance of trees. In addition, it enhances the semantic meaning of pattern trees. The experiments on British Telecom (BT) customer satisfaction dataset show that weighted pattern trees can slightly outperform pattern trees, and both of them are slightly better than fuzzy decision trees in terms of prediction accuracy. In addition, the experiments show that (weighted) pattern trees are robust to over-fitting. Finally, a limitation of pattern trees as revealed via BT dataset analysis is discussed and the research direction is outlined.

Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.