Predict the churn and silent customers: A case study of individual investors

In a typical brokerage firm, most customers are silent or churn investors. However, the prediction of silent investors did not gain enough attention. Based on the CRISP-DM data mining framework and decision tree algorithm, two models are proposed for churn and silent investors respectively. The tree models show that low return rate results in the silent investors, and the fund transfer pattern is of the most importance in both models. Retention strategies are provided based on the behavioral finance theory and the two models´ misclassification rate.