Data Standardization for Enhanced Outcomes in Cluster Analysis Using JMP®
Mantosh Kumar Sarkar, Student, Oklahoma State University
Data transformation plays a critical role in achieving better results from cluster analysis, a popular technique used for market segmentation. Customer survey data used for market segmentation via cluster analysis often exhibits response styles of respondents. Response styles like acquiescence and extreme response styles can bias the results of cluster analysis. Recent research has shown that double-standardization (standardizing across both row and column) and range standardization are good candidates for eliminating response styles from dominating the cluster results, Pagolu et al. (2011). But, JMP does not provide these special standardization features. This paper shows how to achieve double-standardization in JMP by using its powerful scripting language (JSL). To test if our code is working properly, we have used customer survey data from a business-to-business company in hydraulic and pneumatic industry. A 9-point rating scale is used to measure customers’ responses to questions as shown in Fig.1 below.