Mining high utility quantitative association rules

Abstract

Mining weighted association rules considers the profits of items in a transaction database, such that the association rules about important items can be discovered. However, high profit items may not always be high revenue products, since purchased quantities of items would also influence the revenue for the items. This paper considers bom profits and purchased quantities of items to calculate utility for the items. Mining high utility quantitative association rules is to discover that when some items are purchased on some quantities, the other items on some quantities are purchased too, which have high utility. In this paper, we propose a data mining algorithm to find high utility itemsets with purchased quantities, from which high utility quantitative association rules also can be generated. Our algorithm needs not generate candidate itemsets and just need to scan the original database twice. The experimental results show that our algorithm is more efficient than the other algorithms which only discovered high utility association rules. © Springer-Verlag Berlin Heidelberg 2007.