Abstract

This paper describes a method for computing the potential profit of a sales item from cross-selling relationships produced by association rule mining. This method generates a true ranking by which most valuable ite ms are top-ranked as contributing to the increase of total profits even if each item is unprofitable. Such ranking is effective in real situations where some items are loss leaders in daily cross-selling. Unprofitable items in the head of a rule are likely valuable for selling more profitable items. Such potential profit is simply defined and computed in terms of the confidence factors of association rules, thus, efficient and easy implementation is possible. Moreover, presentation by ranking is simple enough to suggest a marketing strategy for sales promotion and advertising.