Association Rules for Purchase Dependency of Grocery Items

Abstract

Customer purchasing behaviour is reflected in the choice of products consumers purchased. An item that a customer purchases sometimes depends on the purchase of another item. Retailers can use purchasing dependencies for planning replenishment of inventory to avoid stock-outs. However, such dependencies are usually not visible. This study uses the data mining approach in finding associations between products purchased by customers from a supermarket and four retail shops. Primary data were obtained from 130 single-sales transactions made over a seven days period by customers of the supermarket and retail stores. Association rules for purchase dependencies were mined using two different algorithms, Apriori and Carma, on IBM SPSS Modeller 15. Results indicated that for retail shops, the purchase of grocery products depends on the availability of fresh food items with 83.33% confidence, and 40% of the customers tend to purchase both items within one transaction. For the supermarket, customers are 27.06% more frequent to buy grocery products together with health beauty products and fresh foods items with 96.66% confidence.
 Keywords: purchase dependency, association rules, apriori model, carma model