To search, Click below search items.


All Published Papers Search Service


A Hybrid GeneticMax Algorithm for Improving the Traditional Genetic Based Approach for Mining Maximal Frequent Item Sets


Mir Md. Jahangir Kabir, Shuxiang Xu, Byeong Ho Kang, Zongyuan Zhao


Vol. 14  No. 10  pp. 27-35


Mining Frequent item sets is one of the most useful data mining methods which discovers important relationships among attributes of data sets. Initially it was developed for market basket analysis, but these days it is used to solve any task where discovering hidden relationships among different attributes is required. Mining frequent item sets plays a vital role for generating association rules, finding correlations and many more interesting relationships among different sort of data. A major challenge in the frequent item set mining task is that it generates a huge number of frequent sub item sets from dense data sets. Researchers proposed mining maximal frequent item sets to overcome this problem. Maximal frequent item sets contain the information of an exponential number of frequent sub item sets since if an item set is frequent each of its sub item sets is also frequent. Very few studies have applied evolutionary algorithms to mine maximal frequent item sets using thorough experimental analysis. In a previous study, we showed the efficiency of using a genetic based approach named GeneticMax to find maximal frequent item sets. In this study we will introduce a new algorithm name, hybrid GeneticMax, which uses local search along with a genetic algorithm to mine maximal frequent item sets from large data sets. The purpose of using the genetic algorithm is that this algorithm based approach is robust and the existing genetic based method which is working fine for a specific problem can be improved by hybridizing it. Experiments are performed on different real world data sets as well as on a synthetic data set. Our new scheme compared favorably to existing GeneticMax under certain conditions which are being evaluated


Association rule mining, Maximal frequent item sets, Genetic Algorithm, Lexicographic tree, Data mining