Performance Analysis of Hybrid Swarm Intelligence Rule Induction Algorithm

Abstract

Data mining is used to extract potential information from data base. Rule induction is used to extract information from data base and display it in IF-THEN rule format. First the classification algorithm builds a predictive model from the training data set and then measure the accuracy of the model by using test data set.This work proposes a hybrid rule induction algorithm using Cooperative Particle Swarm (PSO) with Tabu search (TS), and Ant Colony Optimization (ACO). Real world data base consist of both nominal and continuous attributes. ACO based classification algorithms perform well in nominal data base. PSO based classification algorithms perform well in continuous data base where it converts nominal attributes into numerical values. In conventional PSO, there is no guarantee for local optimal solution. So, the proposed algorithm use tabu search in PSO to improve the search capability and integrate pheromone concept of ACO to handle real world classification problems. It uses cooperative concurrent PSO model to implement the algorithm and run two tasks simultaneously in parallel machines. The output of the work compares with the existing algorithm performance in several public domain data sets. The comparison results provide a evidence that: (a) The proposed algorithm is competitive with existing algorithm with respect to predictive accuracy; and the rule lists discovered by the algorithm are considerably simpler (smaller) than those discovered by the existing algorithm and (b) Reduce the execution time of the algorithm.