AAAI Publications, Twenty-Second International FLAIRS Conference

Font Size: 
Rule Mining and Missing-Value Prediction in the Presence of Data Ambiguities
Kasun Wickramaratna, Miroslav Kubat, Kamal Premaratne, Thanuka Wickramarathne

Last modified: 2009-03-17


The success of knowledge discovery in real-world domains often depends on our ability to handle data imperfections. Here we study this problem in the framework of association mining, seeking to identify frequent itemsets in transactional databases where the presence of some items in a given transaction is unknown. We want to use the frequent itemsets to predict "missing items": based on the partial contents of a shopping cart, predict what else will be added. We describe a technique that addresses this task, and report experiments illustrating its behavior.


Dempster-Shafer belief theory, association rule mining, data imperfections, predicting missing items, itemset tree

Full Text: PDF