Proportion Frequency Occurrence Count with Bat Algorithm (FOCBA) For Rule Optimization and Mining of Proportion Equivalence Fuzzy Constraint Class Association Rules (PEFCARs)

Abstract

Fuzzy Class Association Rules (FCARs) play an important role in decision support systems and have thus been extensively studied. Mining the important rules in FCARs becomes very difficult task, so Enhanced Equivalence Fuzzy Class Rule tree (EEFCR-tree) algorithm is proposed in this work. However, a major weakness of FCARs Miner is that when the number of constrained rules in a given class dominates the total constrained rules; its performance becomes slower than the normal method.  To solve this problem this paper proposes a Proportion of Constraint Class Estimation (PPCE) algorithm for mining Enhanced Proportion Equivalence Fuzzy Constraint Class Association Rules (EPEFCARs) in order to save memory usage, run time and accuracy.  Then, Proportion Frequency Occurrence count with Bat Algorithm (PFOCBA) is proposed for pruning rules which much satisfying the class constraints. Finally, an efficient algorithm is proposed for mining PEFCARs rules. Experimental results show that the proposed EPEFCR-tree algorithm is more efficient than Enhanced Equivalence Fuzzy Class Rule tree (EEFCR-tree), Novel Equivalence Fuzzy Class Rule tree (NECR-tree) Miner results are measured in terms of run time, accuracy and memory usage. Experiments show that the proposed method is faster than existing methods.