Market Basket Analysis by Means of a Growing Neural Network

This paper demonstrates the application of a new growing neural network for market basket analysis. The algorithm to be introduced is an extension of the recently published 'grow when required network' and it is tailored in particular to the processing of large amounts of point of sale scanner data. Two of its most remarkable features are its parsimony regarding the number of parameters to be preset by the user and its considerable autonomy with respect to the network creation process. The simplicity of its implementation and its great flexibility make this algorithm an easy-to-use tool for exploratory data analysis in retailing, marketing and consumer research. The application of the algorithm is demonstrated by means of point of sale scanner data provided by two retail chains.