Apply data mining to segment retail market based on purchasing portfolios

Abstract

Market segmentation is becoming very familiar and essential to every marketer in the process of designing and implementing an effective target-marketing strategy. It is confirmed in the grocery retail industry about the importance of appropriate market segmentation. In this industry, customer purchasing behavior needs to be acknowledged not only in specific products, but also the interaction among the whole range of products. Therefore, the motivation for this thesis is to discover a segmentation based on this purchasing behavior among whole range of products, which is called purchasing pattern. The Purchasing pattern is interpreted by purchasing portfolios, which include list of categories that a certain customer purchases and also consumption behavior on these categories. This thesis is acknowledged from related theories to design a theoretical model of market segmentation based on purchasing portfolios. Then, data mining techniques are applied to process a practical database in order to test the theory’s hypotheses, as well as illustrate for the model. As a result, the availability of segmentation is verified from a technical view and the practical significance of segmentation is confirmed from a marketing view. The result from data mining has shown four segments from the analysis of purchasing portfolios. These four segments cover most of the market, and remain over time. The segmentation is assessed from marketing view to be appropriate for practical application. Furthermore, there are three segments that are selected to be analyzed further. They represent three distinct purchasing behaviors. Three specific purchasing portfolios are built for each segment, which can be used to direct for marketing strategy.