《for B2B Markets》.pdfVIP

SAS Global Forum 2011 Management Paper 193-2011 Comparison of Probabilistic-D and k-Means Clustering in Segment Profiles for B2B Markets Dipanjan Dey, Satish Garla, Goutam Chakraborty, Oklahoma State University, Spears School of Business, Stillwater, OK, U.S. ABSTRACT Customer segmentation is a critical business analysis tool that allows organizations to build customer profiles and plan marketing efforts to satisfy the varying demands of different segments. The objective of the present study is to empirically explore the concept of probabilistic-D clustering for segment profiling in a business-to-business (B2B) market. A SAS® macro that can be used on any data set for application of this technique is developed and reported. To the best of our knowledge, probabilistic-D technique has never been empirically tested in a business environment. It was compared with the widely used k- means clustering technique. Findings indicate a better explanation of customer segment profiles using probabilistic-D clustering because k-means seems to force a high percent of observations in segments where cluster membership probabilities are less than 50% as calculated by probabilistic-D clustering macro. These observations are likely to be non-responsive (or less responsive) to the marketing efforts directed towards respective k-means segments. Usi