Mining association rules at multiple levels helps in finding more specific and relevant knowledge. While computing the number of frequency of an item we need to scan the given database many times. So we used counting inference approach for finding frequent itemsets at each concept levels which reduce the number of scan. In this paper, we purpose a new algorithm LWFT which follow the top-down progressive deepening method and it is based on existing algorithms for finding multiple level association rules. This algorithm is efficient for finding frequent itemsets from large databases.

Multiple-Level Association Rules Counting inference approach Level wise filtered tables Data mining non-uniform support Confidence