Quantitative Association Rule Mining on Weighted Transactional Data

Abstract

In this paper we have proposed an approach for mining quantitative association rules. The aim of association rule mining is to find interesting and useful patterns from the transactional database. Its main application is in market basket analysis to identify patterns of items that are purchased together. Mining simple association rules involves less complexity and considers only the presence or absence of an item in a transaction. Quantitative association mining denotes association with itemsets and their quantities. To find such association rules involving quantity, we partition each item into equi-spaced bins with each bin representing a quantity range. Assuming each bin as a separate bin we proceed with mining and we also take care of reducing redundancies and rules between different bins of the same item. The algorithm is capable in generating association rules more close to real life situations as it considers the strength of presence of each item implicitly in the transactional data. Also the algorithm can be applied directly to real time data repositories to find association rules.