Performance analysis of utility mining algorithms

Utility Mining is one of the recent emerging fields in Data mining. The main objective of utility mining is to discover high utility itemsets from a database. It differs from traditional frequent itemset mining. In frequent itemset mining, frequently occurred items in database are found. But in Utility mining, itemsets with high utility (utility here refers profit, number of items, cost of an item or any user favorite value) are retrieved. Hence utility mining retrieves semantic correlation among items in a database. This helps in better decision making for target markets, cross selling etc. In recent years numerous algorithms like Two-Phase, UP-Growth+, EFIM algorithm, FHM algorithm, HUI-Miner algorithm, IHUP, d2HUP are proposed to find high utility itemsets. These algorithms calculate only positive utility values. There are also algorithms like FHN, HUINIV to calculate high utility with negative values. In this paper performance of FHN and HUINIV are compared and experimental results are discussed.