Churn Prediction

CHURN PREDICTION
There is a data mining application that was developed in an Italian telecommunications institute. A major concern in customer relationship management in telecommunications companiesis the ease with which customers can move to a competitor, a process called “churning”. Churning is a costly process for the company, as it is much cheaper to retain a customer than to acquire a newone. Churn prediction is the task of predicting which types of customers are likely to churn, and more challenging, when they will churn. The task was solved using decision trees which achieved apredictive accuracy of 82%. This good result was only possible due to the introduction of relevant derived features for prediction which were not available in the original data and due to a representationof the data so that temporal aspects could be included. Thus data preprocessing was a key success factor in this application. One interesting aspect of this case study is that it was implementedtwice, based on manual programming on the one hand, and on graphical modeling on the other. This allowed comparing the amounts of work spent by highly paid KDD experts on the application in bothscenarios. As said above, the objectives of the application to be presented here were to find out which types of customers of a telecommunications company are likely to churn, and when. To this end, theavailable data tables were transformed so that classification algorithm could be applied. In the resulting data set, each row (that is, each example for classification) corresponded to one customer of thecompany, and contained many features describing their telecommunication behavior for each of five consecutive months. Whether or not the customer left the company in the sixth month determined theclassification label or target. Thus a binary classification problem was formed that could directly be addressed using several classification algorithms. Once a learned classifier is available it can be...