A Tree-based Fuzzy Average-Utility Mining Algorithm

Utility mining considers high-utility itemsets as useful by combining item quantities and item benefits. Its mining results do not, however, include the quantity information such as large amounts or small amounts. Fuzzy utility mining is thus proposed to fuzzify the results of utility mining and obtain linguistic high-utility itemsets. However, fuzzy utility measurement is not fair to evaluate itemsets because the fuzzy utility value of an itemset in a transaction may be higher than those of its subsets. In the past, we defined the fuzzy average utility mining to solve the above problem and proposed a two-phase method to solve the fuzzy average-utility mining problem and find high fuzzy average-utility itemsets. However, its execution is slow. In this paper, an efficient algorithm is proposed, which uses a tree structure to solve fuzzy average-utility mining. The proposed tree-structure method is compared with the previous two-phase approach. Experimental evaluation shows that the efficiency of the proposed method is better than that of the two-phase algorithm in execution time and numbers of candidates.