A Comparative Study on Transaction Database Mining Algorithms

Abstract

Association Rules is the process of identifying relationships among set of items in transaction database. Finding frequent itemsets is the most expensive step in Association rule discovery. Real world datasets are sparse, dirty and contain hundreds of items. In such situations, discovering interesting rules (results) using traditional frequent itemset mining approach is not appropriate. In this paper, mining association rules for Transaction database using FP-Growth and DynFP-Growth algorithms is presented. FP-Growth algorithm avoids or reduces candidate generation. Moreover, it greatly reduces the need to traverse the database. DynFP-Growth algorithm works in the same process as FP-Growth with lexicographic order, and stores at least one item is detected. Although resulting FPTree is too large to store in memory, it is more flexible with different support values. For the empirical study, three synthetic datasets by IBM Quest data generator is used: (a) T10I4D100K (b) T40I10D30K, and (c) T40I10D100K.