Modeling Large Data Sets in Marketing

In the last two decades, marketing databases have grown significantly in terms of size and richness of available information. The analysis of these databases raises several information-related and statistical issues. We aim at providing an overview of a selection of issues related to the analysis of large data sets. We focus on the two important areas: single source databases and customer transaction databases. We discuss models that have been used to describe customer behavior in these fields. Among the issues discussed are the development of parsimonious models, estimation methods, aggregation of data, data-fusion and the optimization of customer-level profit functions. We conclude that problems related to the analysis of large databases are far from resolved, and will stimulate new research avenues in the near future.