Predicting Missing Items in Shopping Carts

Association mining techniques search for groups of frequently co-occurring items in a market-basket type of data and turn this data into rules. Previous research has focused on how to obtain list of these associations and use these “frequent item sets” for prediction purpose. This paper proposes a technique which uses partial information about the contents of the shopping carts for the prediction of what else the customer is likely to buy. Using Frequent Pattern Tree (FP-Tree) instead of Item set Trees (IT-Tree) and Frequent Pattern Tree (FP-Tree), all the rules whose antecedents contain at least one item from the incomplete shopping cart can be obtained in efficient manner. Rules are then combined and Prediction is done using Bayesian Decision Theory and DS-ARM algorithm based on the Dempster-Shafter theory of evidence combination.