Abstract—Association rule mining for classification is a data mining technique for finding informative patterns from large datasets. Output is in the form of if-then rules containing attribute value combinations in antecedent and class label in the consequent. This method is popular for classification as rules are simple to understand and allow users to look into the factors leading to a specific class label. Rule mining methods based on swarm intelligence, specifically particle swarms, can effectively handle problems with large number of instances and mixed data. But the issue of classification over imbalanced datasets, wherein samples from one class greatly outnumber the other class, has not been fully investigated so far. A rule mining method based on Dynamic Particle Swarm and Ant Colony Optimizer that can handle data imbalance, has been proposed in this paper. Performance of the proposed algorithm has been compared with other state-of-the-art methods. Results indicate that in terms of quality, the proposed method outperforms other state-of-the-art methods.

Index Terms—association rule mining; classification; PSO; Imbalanced Dataset

Cite: Veenu Mangat and Renu Vig, "Intelligent Rule Mining Algorithm for Classification over Imbalanced Data," Journal of Emerging Technologies in Web Intelligence, Vol. 6, No. 3, pp. 373-379, August 2014. doi:10.4304/jetwi.6.3.373-379