This document presents a proposed churn prediction model based on data mining techniques. The model consists of six steps: identifying the problem domain, data selection, investigating the data set, classification, clustering, and utilizing the knowledge gained. The authors apply their model to a data set of 5,000 mobile service customers using data mining tools. They train classification models using decision trees, neural networks, and support vector machines. Customers are classified as churners or non-churners. Churners are then clustered into three groups. The results are interpreted to gain insights into customer retention.