Using an Adapted Classification Based on Associations Algorithm in an Activity-Based Transportation System

Abstract

A lot of research has been carried out in the past by using association rules to build more accurate classifiers. The idea behind these integrated approaches is to focus on a limited subset of association rules. However, these integration approaches have not been tested yet within the context of transportation research. The aim of this chapter is therefore to evaluate the performance of an adapted well-known associative classification algorithm on the datasets that are used in the Albatross transportation modelling system. The presented work is an extension of previous research efforts in the sense that it now becomes possible to use the adapted CBA system for multi-class problems. Experiments showed that the original CBA system achieved the best average performance for the three classifiers under evaluation. While the adapted CBA still generated better average results than CHAID, the performance with respect to original CBA was slightly worse.