Author(s)

Li Shen, Hong Shen, Paul Pritchard & Rodney Topor

Abstract

The largest itemset in a given collection of transactions £> is the itemset that occurs most frequently in T>. This paper studies the problem of finding the A/" largest itemsets, whose solution can be used to generate an appropri- ate number of interesting itemsets for mining association rules. We present an efficient algorithm for finding the jV largest itemsets. The algorithm is implemented and compared with the naive solution using the Apriori approach. We present experimental results as well as theoretical analysis showing that our algorithm has a much better performance than the naive solution. We also analyze the cost of our algorithm and observe that it has a polynomial time complexity in most cases of practical applications. 1

Keywords