Optimising financial forecasting: implementing a predictive cash flow platform for bank branches

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.