ACCF: Associative Classification Based on Closed Frequent Itemsets

Recent studies in data mining have proposed a new classification approach, called associative classification, which, according to several reports, such as , achieves higher classification accuracy than traditional classification approaches such as C4.5. However, the approach also suffers from one major deficiency: a training data set often generates a huge set of rules. It is challenging to store, retrieve, prune and sort a large number of rules efficiently for classification, especially on dense databases. In this study, we propose a new associative classification method, ACCF(associative classification based on closed frequent itemsets). The method extends an efficient closed frequent pattern mining method, Charm to mine all frequent closed itemsets (CFIs) and their tidsets, which would help to generate the class association rules (CARs). And we also adopt a new way to classify an unseen case correspondingly. Our extensive experiments on 18 databases from UCI machine learning database repository show that ACCF is consistent, highly effective at classification of various kinds of databases and has better average classification accuracy in comparison with CBA. Moreover, our performance study shows that the method helps to solve a number of problems that exist in the current classification systems.