Performance analysis of hybrid rule Induction algorithm using swarm Intelligence techniques

Abstract

With the enormous amount of data stored in files databases and newlineother repositories it is important to develop a powerful analysis tool to extract newlineinteresting knowledge from data that could help in decision making Data newlinemining is a process that uses a variety of data analysis tools to discover newlinepatterns and relationships in data Data mining is defined as The nontrivial newlineextraction of implicit previously unknown and potentially useful information newlinefrom data newlineRule induction is an area of machine learning in which formal rules newlineare extracted from a set of observations The performance of the classification newlinemethods are estimated and evaluated according to predictive accuracy speed newlinerobustness scalability and interpretability This research concentrates on the newlineperformance enhancement of rule induction algorithm through hybridizing the newlineswarm intelligence techniques and implements it by using cooperative newlinecoevolution framework newlineAnt Colony Optimization ACO and Particle Swarm Optimization newline PSO are the two important swarm intelligence techniques used in data newlineMining From the literature review it is observed that the existing ACO based newlinerule induction algorithms support only nominal attributes It uses newlinediscretization technique to convert continuous attributes into nominal newlineattributes PSO based rule induction algorithms support only continuous newlineattributes and use indexing techniques to represent nominal attributes newlineEvolutionary algorithms require an encoding scheme to represent the newlineAttributes Tabu search is a powerful stochastic optimization technique The newlineincorporation of tabu search TS as a local improvement procedure helps to newlineexplore the search space efficiently Sequential based approach algorithm newlinetakes more time to discover a rule set from a large dataset Cooperative newlinecoevloution frame work is used to solve large scale problems newline