Study On The Application Of Multi Expression Programming In The Rule Mining

Data mining is a frontier subject of many subjects cross and fusion,it is one of theimportant achievements in the process of information technology development, nowadaysits theoretical research and practical application have been concerned widely. Associationrule mining is an important research task in data mining, it is used to find theimplicit association among a large set of data items. In recent years, with the rapidpopularization of the Internet and the continuous expansion of the size of thedatabase, association rule mining is becoming more and more complex, the traditionalassociation rule mining algorithms cannot meet the needs of data mining.Many scholarshave begun to apply many kinds of algorithms to solve the problem of datamining, evolutionary computation is a very effective method and it has become one of thefocus of attention.Gene expression programming is a new model of evolutionary computation, it mainlybased on the way of gene expression in biological evolution.Multi ExpressionProgramming is a new extension of the field of Genetic Programming. In recent years,because of its unique and flexible encoding method, Multi Expression Programming hasbeen widely applied in many fields, and also achieved certain results.This paper mainly studies the Multi Expression Programming algorithm and use it tosolve the problem about association rule mining. Firstly, mainly studies basic principles andtechniques of the Gene Expression Programming and Multi Expression Programming.Following detailly analyzes the advantages and disadvantages of the classical associationrule mining algorithm. Finally, aiming at the problem of association rule mining, designsand improves of Multi Expression Programming algorithm, so that it can solve the problemof association rule mining more better.In this paper, the main work and innovations are summarized as follows:(1) The genetic operators and fitness function of Multi Expression Programmingalgorithm have been designed.The association rule mining of data mining has beenstudied,the traditional association rule mining algorithm has been analyzed detailly. Andcombining with the question of association rule mining、 the coding way of Multi Expression Programming and the advantages and disadvantages of Multi ExpressionProgramming algorithm,fitness function and several different genetic operators.(2) The improvements about Multi Expression Programming algorithm. In order tomake better use of the advantages,the better to make up for its shortcomings, put forwardthe improved Multi Gene Expression Programming、 the combined algorithm of MultiExpression Programming and differential evolution algorithm、the combined algorithm ofMulti Expression Programming and simulated annealing algorithm. And these algorithmsare applied to solve the problem of association rule mining.(3) The experiment of algorithms. Through the analysis of the experimental resultsprove the feasibility of the algorithm for solving the association rule mining problem.Andresults showed a lot of advantages.