An Approach of Frequent Item Tree for Association Generation

K. Amphawan and A. Surarerks (Thailand)

Keywords

FPgrowth algorithm, frequent itemsets, candidate itemsets, apriori algorithm

Abstract

An important problem in the data mining field is a frequent pattern mining problem. Efficient methods for mining frequent patterns have been studied extensively by many researchers. Frequent itemset generation is the most time-consuming process in association rules mining. One interesting technique for generating frequent itemsets using frequent pattern tree (FP-tree) is FP-growth algorithm. This algorithm is a valuable remark for various further developments. In this paper, we propose an algorithm for creating a new FP-tree called frequent item tree. It is shown that the complete set of frequent itemsets can be generated using a single tree. We also propose an algorithm for mining frequent itemsets from frequent item tree. In an experimental evaluation of our algorithm on synthetic data shows the advantage of frequent item tree over a FP-growth algorithm, in terms of runtime.

Important Links:



Go Back