Abstract:
This study addresses the challenge of managing daily cash flows in bank branches, which often face excess or deficiency of cash, disrupting daily operations. The research focuses on developing a predictive model that can accurately forecast daily cash inflows and outflows across 175 bank branches in Sri Lanka, covering all provinces and districts. The aim is to create a robust tool that enhances financial efficiency by reducing idle cash balances while ensuring smooth operations. Two models were developed for this purpose: a multi-branch model utilizing Artificial Neural Networks (ANN) and a single-branch model using a Random Forest Regressor. The multi-branch model, which features separate sub-models for cash inflow and outflow, attained accuracies of 85.45% and 85.50%, respectively. In contrast, the single-branch model, tested on the Grandpass branch, demonstrated performance with accuracies of 60.06% for cash inflow and 70.29% for cash outflow predictions. The multi-branch model's superior performance underscores its ability to provide consistent and reliable predictions across a broader range of branches. The final models have been integrated into a web-based user interface, offering a user-friendly platform for real-time cash flow predictions. Overall, the results highlight the multi-branch model as a robust solution for effective cash flow management across bank branches.
