A New Associative Classifier based on CFP-Growth++ Algorithm

Abstract

Association rule mining and Classification are the important task in data mining. The mining of frequent patterns with single minsup-based frequent pattern mining algorithms such as Apriori and FP-Growth leads to rare item problem. So to overcome this we use multiple minsup based frequent pattern mining algorithm called CFP-Growth++ for extraction of frequent patterns. The association rule mining is carried out to extract the rules and then the rule based classifier is build. In this, we have proposed a new associative classification method called Rule based classifier using CFP-Growth++. The method extends the efficient Frequent Pattern Mining method called CFP-Growth++ and constructs the MIS tree to mine the large datasets. The rules are generated from the frequent patterns mined from CFP-Growth++ and rule-based classifier is build. Our extensive experiments on datasets from UCI machine learning database repository show that our new technique is consistent, highly effective at classification of various kinds of datasets and has better average classification accuracy in comparison with FP-Growth based classifier. Moreover, our performance study shows that the method is highly efficient and scalable in comparison with previous associative classification methods.