| Abstrakt: |
Customers churn is an important issue that is always concerned by banks, and is put at the forefront of the bank's policies. The fact that banks can identify customers who are intending to leave the service can help banks promptly make policies to retain customers. In this paper, we propose a combined deep learning network models to predict customers leaving or staying at the bank. The proposed model consists of two levels, Level 0 consists of three basic models using three Deep Learning Neural Networks, and Level 1 is a logistic regression model. The proposed model has obtained evaluation results with accuracy metrics of 96.60%, precision metrics of 90.26%, recall metrics of 91.91% and F1 score of 91.07% on the dataset "Bank Customer Churn Prediction". [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink