ESTIMATION OF PRODUCT SUBSTITUTABILITY AND COMPLEMENTARITY

Deep knowledge of product relationships represents a strategic advantage for retailers. This work proposes a methodology for classifying relationships between two products based on transactional data. Firstly, data mining framework }alled association rules is used for identification of substitutes and complements. The next step is verification of the product pair relationship via estimation of products' price cross-elasticity. The methodology was applied on real dataset with retailer's transactions. The empirical results clearly show that stand-alone usage of association rules provides only pure classification for product pairs while incorporation of product price cross-elasticity helps to increase the accuracy of product pairs categorization.