Providing a customer churn prediction model Using Random Forest technique

In order to succeed in the global competition, organizations need to understand and monitor customers' behavior, so that they could retain them by predicting their preference and behavior before others. Recently, marketing strategies have been changed from product-oriented strategies to customer-oriented strategies and most organizations have focused on customer relationship management. In fact, more organizations have found out that retention of their present customers, as their most valuable asset, is very important. Therefore, with the aim of describing data mining abilities in churn management, and designing and implementation of a customer churn prediction model using a standard CRISP-DM (Cross Industry Standard Process for Data Mining) methodology based on RFM (Recency, Frequency, Monetary) and random forest technique, the database of one of the biggest holdings of the country, Solico food industries group, is explored. Using this model, the customers tending to turn over are identified and effective marketing strategies will be planned for this group. Customer behavior analysis indicates that length of relationship, the relative frequency and the average inter purchase time are among the best predictors.