Mining high-utility itemsets with irregular occurrence
Supachai Laoviboon, Komate Amphawan · 2017
High-utility itemsets mining (HUIM) is proposed to discover itemsets giving high utilities (such as high profit, low cost/risk and other factors). This can help to extract hidden-knowledge from buying behavior of customers. However, HUIM may not sufficiently give hidden-knowledge and observe occurrence behavior of itemsets in some applications, since it only considers utilities of items/itemsets. Thus, we propose to mine high utility itemsets with irregular occurrence (also called High Utility-Irregular Itemsets, HUIIs). HUIIs can help to gain knowledge about “products giving high profits even if customers do not regularly purchase them together” and to improve marketing strategies and sale profit. To mine HUIIs, an efficient single-pass algorithm based on the use of new modified utility-list structure, called HUIIM (HUIIs-Miner), is designed. Experiments on real and synthetic datasets were done to investigate computational time and memory consumption of HUIIM.