Knowledge discovery of hidden consumer purchase behaviour: a market basket analysis

Abstract

In this study, we aim to apply data mining techniques to uncover hidden patterns in consumer behaviour. We use a cross-sectional data representing a random sample of consumers in Kuwait. A novel data mining technique known as market basket analysis, which was first introduced to the marketing research literature in the 1990s, is applied for the first time to investigate consumer behaviour in Kuwait. This technique is also known in the literature as association mining or affinity analysis as it attempts to extract association rules by determining which sets of products or items tend to purchased together. We also use multidimensional scaling technique in order to visualise perceived similarities among various products. Our results indicate that significantly better decisions might be made by discovering and visualising hidden patterns in readily-available supermarket transactional data. We believe that knowledge discovered might be used in making better supermarket layout decisions. In fact once association rules are discovered, MBA results might be applied in several scenarios, including market segmentation, designing catalogues for up-selling or cross-selling, and designing promotion and advertising strategies.