Interactive Graphical Methods in the Analysis of Customer Panel Data

Abstract

The study of complex market phenomena is often supported by data gathered from panels of customer households. Recent advances in statistical graphics research help analysts do justice to the richness of such data. We describe a graphical approach based on the use of simple linked plots in which case identity is preserved. The expressive power of these plots is enhanced by a variety of interactions such as painting and identification. We also introduce the case-profile plot, developed to support the exploratory analysis of panel data. This approach is illustrated by examples from packaged-goods retailing and telecommunications.

KEY WORDS:

• Customer response models
• Multivariate data
• Visualization