Discovering Maximal Frequent Item set using Association Array and Depth First Search Procedure with Effective Pruning Mechanisms

The first step of association rule mining is finding out all frequent itemsets. Generation of reliable association rules are based on all frequent itemsets found in the first step. Obtaining all frequent itemsets in a large database leads the overall performance in the association rule mining. In this paper, an efficient method for discovering the maximal frequent itemsets is proposed. This method employs Association array technique and depth first search technique to mine Maximal Frequent Itemset. The proposed algorithm GenMFI takes vertical tidset representation of the database and removes all the non-maximal frequent item-sets to get exact set of MFI directly. Pruning is done for both search space reduction and minimizing the number of frequency computations and number of maximal frequent candidate sets. The algorithm gives better results for the sparse dataset even though number of the Maximal Frequent Itemset is huge. The proposed approach has been compared with Pincer search algorithm for T10I4D100K dataset and the results shows that the proposed algorithm performs better and generates maximal frequent patterns faster. In order to understand the algorithm easily, an example is provided in detail.