Affiliation:
1. IDRBT, India
2. Institute for Development and Research in Banking Technology, India
Abstract
In this chapter, we model association rule mining as a Fuzzy multi-objective global optimization problem by considering several measures of strength such as support, confidence, coverage, comprehensibility, leverage, interestingness, lift and conviction by utilizing various fuzzy aggregator operators. In this, pdel, each measure has its own level of significance. Three fuzzy multi-objective association rule mining techniques viz., Fuzzy Multi-objective Binary Particle Swarm Optimization based association rule miner (FMO-BPSO), a hybridized Fuzzy Multi-objective Binary Firefly Optimization and Threshold Accepting based association rule miner (FMO-BFFOTA), hybridized Fuzzy Multi-objective Binary Particle Swarm Optimization and Threshold Accepting based association rule miner (FMO-BPSOTA) have been proposed. These three algorithms have been tested on various datasets such as book, food, bank, grocery, click stream and bakery datasets along with three fuzzy aggregate operators. From these experiments, we can conclude that Fuzzy-And outperforms all the other operators.
Reference42 articles.
1. Mining association rules between sets of items in large databases
2. Fast algorithms for mining association rules.;R.Agrawal;Proceedings of the 20th International Conference on VLDB Conference,1994
3. MODENAR: Multi-objective differential evolution algorithm for mining numeric association rules
4. Anandhavalli, M., Kumar, S.S., Kumar, A., & Ghose, M. K. (2009). Optimized association rule mining using genetic algorithm. Advances in Information Mining, 1-4.
5. New binary PSO based method for finding best thresholds in association rule mining.;A.Asadi;Applied Soft Computing,2012