A non-recursive approach for FP-tree based frequent pattern generation

Generation of frequent patterns is an important component of association rule mining. FP-tree and P-tree are two efficient algorithms for the purpose, requiring two and one scan of database to generate the compact data structures Frequent Pattern-tree and Pattern-tree, respectively. P-tree algorithm restructures the P-tree and creates FP-tree. Both the algorithms further, use a recursive mining process to generate frequent patterns from FP-tree. In this paper, a new approach is bottom-up scanning. Next, it uses a highly efficient non-recursive procedure to generate frequent patterns. This new approach has been compared with two algorithms FP-growth and P-tree and has been found to be better in terms of performance.