| Summary: | 碩士 === 國立中央大學 === 工業管理研究所 === 92 === Cross-level association rules mining with multiple minimum supports is an important generalization of the association rule mining problem. Instead of setting a single minimum support for all items, Liu et al. proposed a method, named MSApriori, to allow users for specifying multiple minimum supports to reflect the natures of the items. Because not all items are sold in a whole year, we should consider the transactions in the sold periods of items as we calculate the supports of items. Previous techniques for mining cross-level association rules with multiple minimum supports are most top-down, progressive depending method extended from Apriori algorithm, e.g. MMS_Cumulate. Previous approaches result in worse mining efficiency and incompleteness of mined rules.
In this research, we propose a bottom-up, simultaneously merging method based on CL_FP-tree, called CL_FP-tree (MIS), to improve the mining efficiency and completeness of mining cross-level association rule with multiple minimum supports. We extend the procedure, which proposed by Alex H.W. Lin (2003) for supports counting, not only to count the supports for all items, but also to judge the sold periods of all items as the basis of support counting. CL_FP-tree (MIS) aims to reduce the number of database rescans for finding the cross-level information.
We implement the CL_FP_tree (MIS) with real data and find the results that the efficiency of CL_FP_tree (MIS) is better than CL_Apriori(MIS). And the number of cross-level association rules found by CL_FP-tree (MIS) algorithm is more than CL_Apripri(MIS). Furthermore, we solve the problem that the number of 1-frequent items decreasing as long as the number of transactions increasing.
|
|---|
