A Comparative Study of Apriori and Rough Classifier for Data Mining

Abstract

This paper presents a comparative study of two data mining techniques; apriori AC and rough classifier Rc. Apriori is a technique for mining association rules while rough set is one of the leading data mining techniques for classification. For the classification purpose, the apriori algorithm was modified in order to play its role as a classifier. The new apriori called AC is obtained through the modification of the frequent item set generation function and a filtering function is proposed. The purpose of this modification is to consider the apriori as a target oriented training where target class is included during mining. Frequent item set generation phase is carried out to mine all attributes together with target class. The performance of AC is compared with a rough classifier. Rough classifier RC is chosen for comparison for its rule based structure. Three important measures will be used for both techniques, the accuracy of classification, the number of rules, and the length of rules. The experimental result shows that AC is comparable with RC in terms of accuracy and in several experiments it performs better. AC. produced more rules than RC. This study indicates that apriori can be used as an alternative classifier.