A Comparative Analysis of Breadth First Search Approach in Mining Frequent Itemsets

The ensuing paper contrasts the algorithms using breadth first search technique of mining frequent itemset. The Apriori algorithm is a well-known algorithm which adopts breadth first search approach to mine frequent itemsets. The process of finding the frequent itemsets is simple for smaller database and complex for larger databases. Candidate itemset generation is a major disadvantage for Apriori algorithm. In this research paper, the proposed BFS-LP-Growth algorithm is compared with another breadth first search algorithm Apriori. The proposed BFS-LP-Growth algorithm overcomes the limitation of Apriori algorithm by processing the database twice and also avoids the exploration of candidate itemset. The structure of LP-Growth algorithm makes the proposed BFS-LP-Growth algorithm more advantageous. The breadth first search enhances the performance of the proposed BFS-LP-Growth algorithm by creating subheader table which avoids the necessity of conditional pattern base and conditional LP-tree. Standard databases used for comparison are Chess and Mushroom. Based on the results, we find that the proposed BFS-LP-Growth algorithm is more worthwhile than the Apriori algorithms as execution time is low and the need for candidate patterns is diminished.