Improved Maximal Length Frequent Item Set Mining
D. Haritha, Vedula Venkateswara Rao · 2012
Association rule mining is one of the most important techniques in data mining. Which wide range of applications It aims it searching for intersecting relationships among items in large data sets and discovers association rules. The important of association rule mining is increasing with the demand of finding frequent patterns from large data sources. The exploitation of frequent item set has been restricted by the large number of generated frequent item set and high computational cost in real world applications. To avoid these problems we can use maximum length frequent item sets in generating association rules. The maximum length frequent item sets can be efficiently discovered on very large data sets. At present in research we have LFIMiner algorithm and MaxLFI algorithm to generate maximum length frequent item sets. Here we are proposing a new algorithm called FPMAX for generating maximum length frequent item sets that uses lattice graph data structure. itemsets (FI) and closed frequent itemsets (FCI). MAFIA assumes that the entire database (and all data structures used for the algorithm) completely fit into main memory. Since all algorithms for finding association rules, including algorithms that work with disk-resident databases, are CPU-bound, we believe that our study sheds light on some important performance bottlenecks. In a thorough experimental evaluation, we first quantify the effect of each individual pruning component on the performance of MAFIA. Because of our strong pruning mechanisms, MAFIA performs best on dense datasets where large sub trees can be removed from the search space. On shallow datasets, MAFIA is competitive though not always the fastest algorithm. On dense datasets, our results indicate that MAFIA outperforms other algorithms by a factor of three to thirty. At present, LFIMiner ALL is the fastest algorithm for mining maximum length frequent itemsets. Exploiting the optimization techniques in LFIMiner ALL algorithm, we develop the FPMax algorithm to discover maximum length frequent itemsets by adding Lattice Graph maximum length frequent itemsets to prune the search space. Experimental results on real-world datasets show that our proposed algorithm is faster than LFIMiner algorithm for mining maximum length frequent itemsets. The rest of the paper is organized as follows. Sect. II provides Literature Survey of frequent Item Sets. The System Design is presented in Sect. III. The Proposed Algorithm is presented in Sect. IV, whereas in Sect. V we discuss experimental results. Several final remarks and a brief discussion on future work conclude in Section VI.