A Review of High Utility Itemset Mining for Transactional Database

High utility itemset mining (HUIM) is an expansion of frequent itemset mining (FIM). Both of them are techniques to find interesting patterns from the database. The interesting patterns found by FIM are based on frequently appeared items. This approach is not that efficient to identify the desired patterns, as it considers only existence or nonexistence of items in database and ignores utility. However, the patterns are more meaningful for the user if the utility is considered. The utility can be quantity, profit, cost, risk, or other factors based on user interest. HUIM is another approach to find interesting patterns by considering utility of items along with the frequency. It uses minimum utility threshold to determine if an itemset is high utility itemset (HUI) or not. There are several challenges to implement utility from traditional pattern mining to HUIM. Lately, there are many research contributions that proposed different algorithms to solve these issues. This review work explores various HUIM techniques with detailed analysis of different strategies like apriori, tree based, utility lists based, and hybrid. These strategies are used to implement various HUIM techniques in order to achieve the effectiveness in pattern mining. The observations and analytical findings based on this detailed review done with respect to various parameters can be recommended and used for further research in the pattern mining.