Web data mining to monitoring marketing performance. Focus on potential customers risk of churn

This PhD thesis shows the strategic magnitude of the predictive data mining models in today competitive landscape for discovering hidden knowledge collected in huge databases in order to maximize the probability of customer conversion and minimize their risk of churn. The main challenge for decision makers is to discover those customer are likely to churn. In particular, the attention has been paid on the main data mining techniques helpful to forecast the potential customers risk of churn within global organizations with an outside-in perspective in the web marketing field. The database analyzed was provided by a global company which develop web analytics services all over the world.