A Novel Algorithm for Mining Fuzzy High Utility Itemsets

Abstract

Utility mining is to find the itemsets in a transaction database with high utility values like profits. The proposed algorithms do not reflect the fuzzy degree of purchased quantity and profit level for mined high utility itemsets, which are essential fo r decision making in various applications like stock control and sales analysis. In this paper, we explo re to apply fuzzy sets theory to the utility mining probl em and propose a novel method, namely FHUI (Fuzzy High Utility Itemsets)-Mine , for mining fuzzy high utility itemsets. In addition to reflect the fuzzy degree for quantity and profit regions of high utility ite msets, FHUI-Mine also provides a fuzzy threshold range that may include itemsets with profits slightly less tha n the designated threshold value. To prove the feasibilit y of, FHUI-Mine , it was compared with the well-known Two-Phase algorithm through experimental evaluation. The results show that FHUI-Mine delivers higher mining capability since it can not only mine all hi gh utility itemsets found by Two-Phase algorithm but also discover additionally itemsets that are potent ially high utility ones.