Enhanced Customer Churn Prediction in the Banking Sector Using Random Forest

With increasing competition among banks and the rising expectations of Generation Z customers for digital services, predicting customer churn has become crucial. On the other hand, machine learning models are evolving, and the Random Forest algorithm stands out for its effectiveness in this area. In our research, we preprocessed and visualized data to enhance the quality of the input data and to gain initial useful insights. We carefully selected key features to speed up the algorithm and used strategies to balance the imbalanced data. Implementing the Random Forest model, we achieved an accuracy of about ۹۶%, successfully identifying both loyal customers and potential churners. Finally, we optimized the model's performance with Grid Search CV and Randomized Search CV, enhancing its effectiveness.

کلیدواژه ها:

Customer churn prediction ، Banking sector ، Random forest ، Hyper-parameters Tuning ، Grid Search Cross Validation ، Randomized Search Cross Validation