Rup/Frup-growth: An Efficient Algorithm for Mining High Utility Itemsets

Abstract

High Utility Itemsets Mining is a topic in the field of data mining. Many algorithms have been proposed in recent years, but they generate a large number of candidate itemsets for high utility itemsets. So the bottleneck of these algorithms is the processing of the candidates; the situation may become worse with the increasing of the number of long transaction itemsets or the decreasing of the minimum utility threshold. Meanwhile, in real-life marketing, the profit value of an item should also be considered along with its selling frequency and quantity. To address these issues, proposes an improvement of the UP-Growth algorithm called RUP-Growth, and develops a new algorithm called FRUP-Growth to take into consideration both the minimum support number and the minimum utility value to mine frequent & high utility itemsets. The experimental results show that our proposed strategies are more efficient and effective; especially with real-life marketing database, the advantage is more obvious.