<![CDATA[Deposits serve as the primary source of funding for banks and play a crucial role in determining their ability to extend loans. Therefore, banks must ensure that deposit levels remain stable, one of which can be achieved by increasing customer interest through effective marketing strategies. Telemarketing is considered an important channel for promoting the deposit program. Predictive analysis was believed can enhance and optimize marketing strategies. Previous studies have predominantly focused on improving accuracy metrics without emphasizing their interpretability. This study aims to develop a machine learning model that employs appropriate evaluation metrics while also providing insights into the factors influencing customers' decisions to make deposits. Several machine learning algorithms are applied in this research, including Decision Tree, Random Forest, and XGBoost. Experimental results indicate that the XGBoost algorithm combined with a random undersampling approach achieved the best performance with a recall score of 67% and an accuracy of 94% in predicting uninterested customers. Additionally, SHAP value analysis is utilized to identify the contribution level of each variable to customers’ deposit decisions. The interpretation results suggest that banks should pay attention to the choice of communication media, the appropriate timing for contacting customers, and the amount of balance owned by each customer.]]>
Copyrights © 2025
