An Innovative Approach for finding Frequent Item sets using Maximal Apriori and Fusion Process and Its Evaluation

Abstract

Frequent pattern mining is a vital branch of Data Mining that supports frequent itemsets, frequent sequence and frequent structure mining. Our approach is regarding frequent itemsets mining. Frequent item sets mining plays an important role in association rules mining. Many algorithms have been developed for finding frequent item sets in very large transaction databases. This paper proposes an efficient SortRecursiveMine (Sorted and Recursive Mine) Algorithm for finding frequent item sets. This proposed method reduces the number of scans in the database by first finding the maximal frequent itemsets in the database and then all its subset consider as frequent according to Apriori property. Then reduce the database by just considering only those transactions which are 1-Itemset frequent but not contain in frequent itemsets and then mine the remaining left frequent itemsets. Our proposed SortRecursiveMine algorithm works well based on recursive condition. Thus it reduces the memory constraints and helps to efficiently mine frequent itemsets in less time. At last we are evaluating this method, and performed an experiment on a real dataset to test the run time of our proposed algorithm.