Performance Evaluation of Fuzzy Association Rule Mining Algorithms

In the field of data mining, interesting rules are generated based on manifold criteria for the classification of itemsets. Association rule mining is used to generate rules based on support, confidence, lift, leverage, etc. by following three techniques - Boolean, Quantitative and Fuzzy association rule mining. Boolean and Quantitative association rule mining techniques work with crisp value and quantitative numeric value respectively for which many interesting rules are being lacked. On the other hand, Fuzzy association rule mining uses linguistic values and membership functions to determine the rules and so it provides more realistic values from the other two association rule mining techniques. Instead of using crisp and quantitative values for rule mining, Fuzzy logic system with Evolutionary algorithm framework can resolve this lacking. The Fuzzy logic system, adapted by Evolutionary Genetic Algorithm which is a global search heuristic technique that optimizes rule generation. In this paper, the performance of Fuzzy association rule mining algorithms is compared. For performance analysis, classical Apriori algorithm, FuzzyApriori algorithm, and Evolutionary GeneticFuzzyAprioriDC algorithm are being used.