Differential Evolution and Enhanced Distributed Fuzzy Associative Classifier for Big Data

Abstract

A latest and potential methodology incorporating classification and association rule mining is referred as Associative classification to design a prediction model and attains superior accuracy. Classification with base on  rules in association is regarded to be efficient and beneficial in several scenarios. But, there exist a problem so-known as "sharp boundary" in association rules mining having quantitative attribute domains. The important reason behind this is that learning of fuzzy Associative Classifiers (ACs) is an extremely tedious task, particularly when working with huge datasets. In this research work, an associative classification scheme, called as Classification with Fuzzy Association Rules is proposed, where the fuzzy logic is helpful in the partitioning of the domains for accuracy improvement. The learning algorithm at first helps in the mining of a bunch classification rules in association of fuzzy via extension in fuzzy with segregated variant from popular algorithm of Frequent Pattern growth that depends on fuzzy entropy. Also for improving the system performance, selection model for rule of cleaning optimally to base of proposal in machine learning. Optimal cleaning rule selection is implemented after cleaning rule performance evaluation system is defined. A probable solution that could adopt Differential Evolution (DE) has been introduced for choosing the rules and criteria from an initial rule base.