Improving associative classification by incorporating novel interestingness measures

Associative classification has aroused significant attention in recent years and proved to be intuitive and effective in many cases. This paper aims at achieving more effective associative classifiers by incorporating two novel interesting measures, i.e. intensity of implication and dilated chi-square. The former is proposed in the beginning for mining meaningful association rules and the latter is designed by us to reveal the interdependence between condition and class variables. Each of these two measures is applied, instead of confidence, as the primary sorting criterion under the framework of the well-known CBA algorithm in order to organize the rule sets in a more reasonable sequence. Three credit scoring datasets were applied to testify our new algorithms, along with original CBA, C4.5 decision tree and neural network as benchmarking. The results showed that our algorithms could empirically generate accurate and more compact decision lists