Mining of High Value Item Sets with Negative Utility

Abstract

The patterns generated by frequent pattern mining aims to find the frequent items without considering the utilities of the different items. The traditional association rule mining treats all items to be of equal utility. This is not always the case for a real world application. Utility based data mining is a new area of research and is complementing the frequency based approach. The main objective of Utility Mining is to identify the item sets with highest utilities, by considering profit, quantity, cost or other user preferences as the Utility of the item. Recent approaches developed so far considers the utilities of items to be same over a particular period of time. In our approach we have proposed that the utility of items vary over a period of time. Our work also proposed that the utility of items may also assume negative values. Our work thus treats the data mining in more realistic manner