Advanced classification and time-series methods in marketing.

This collection of essays investigates diverse marketing issues ranging from customer retention to the elaboration of pan-European strategies, t hrough the use of advanced modeling techniques. This work follows two ma in streams of research, from the methodological viewpoint, but also with respect to the issues envisioned. As far as the methodology is concerne d, the first part concerns the development and the application of novel binary choice models that conceptually originated in the statistical mac hine learning literature. In turn, the second part is based on time-seri es analysis. In particular, we devote a special attention to the concept of Granger causality in the time and frequency domains. In addition to this unidirectional concept, we also use a measure of co-movement betwee n multiple time-series. Both parts also differ with respect to the issues envisioned. The main i nterest in the first part is customer retention and churn prediction. In the second part, we focus on the understanding of the similarities and divergences, as well as the mechanisms of interaction between the Europe an countries. In particular, we assess the predictive content of Busines s Tendency Surveys published by the European Commission, allowing for cr oss-country effects. In a second stage, we decompose this predictive con tent over the frequency band. Finally, we investigate to what extent the European countries exhibit homogeneous consumer confidence for various planning horizons and geographic regions. ------------------------------------------------------------------------ --------------------------status: publishe