The Research of A-Priori Algorithm Candidates Based on Support Counts

Abstract

Apriori algorithm is one of the most standard, popular way among associational rule mining algorithms. Its famous character 'all the sub-frequent itemsets are frequent itemsets, and no one itemset is frequent if its sub-frequent is infrequent itemsets' has been reduction the number of candidate itemsets, but the number of candidate itemsets is still large. Moreover, the ratio is so small between frequent itemsets and candidate itemsets that the largest time is wasted in checking the candidate itemsets. According to the character of Apriori, this paper sums up three regularities to improve the ratio of useful candidates. At the same time the independent support count is cited to prune the frequent itemsets who are independent their father itemsets. The optimize algorithm has been proved and realized by one database. This paper is arranged as the following: section 2 presents some key definitions on these rules such as support count, attenuation to support count, constant support count on fixed itemsets and three rules will be introduced shortly. Demonstrations and applications on these rules will be detailed in section 3. Section 4 introduces the algorithm of rules and an example is introduced to test the efficiency of these rules. The last section draws some conclusions and future works.