The prediction of customer churn is crucial for any firm, but particularly for a banking firm as it is ordinarily cheaper to maintain an existing customer than obtain a new one. This paper aims at the use of Artificial Neural Networks (ANN) in modeling customer churn with analysis based on both static and dynamic characteristics. Criticizing prevalent models that depend on instantaneous data, this study incorporates time series data to enhance predictive precision. To control class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) were used in improving the performance of the models. Compared to Decision Tree, Random Forest, XGBoost and SVM, the proposed ANN gives the highest accuracy of 89.5%, precision of 91.0%, and F1-score of 89.2%. The outcomes of this study imply that the ANN model, with its desired low error rate, is suitable for identifying customers planning to churn in a timely manner and provides recommendations for more efficient anti churn measures.