Mining Interesting Infrequent and Frequent Itemsets Based on Minimum Correlation Strength

IMLMS (interesting MLMS (Multiple Level Minimum Supports)) model, which was proposed in our previous works, is designed for pruning uninteresting infrequent and frequent itemsets discovered by MLMS model. One of the pruning measures used in IMLMS model, interest, can be described as follows: to two disjoint itemsets A,B, if interest(A,B)=|s(A∪B) s(A)s(B)|<mi, then A∪B is recognized as uninteresting itemset and is pruned, where s(⋅) is the support and mi a minimum interestingness threshold. This measure, however, is a bit difficult for users to set the value mi because interest (A,B) highly depends on the values of s(⋅). So in this paper, we propose a new measure, MCS (minimum correlation strength) as a substitute. MCS, which is based on correlation coefficient, has better performance than interest and it is very easy for users to set its value. The theoretical analysis and experimental results show the validity of the new measure.