A Survey on Cash Demand Forecasting for ATM's Using Different Financial Modelling Techniques

Abstract

Optimization of cash management techniques for automated teller machines pose an arduous task due its unpredictable nature of cash withdrawal pattern throughout the year. This uncertainty is caused by several factors such as weekends, salary days, holidays etc. resulting in non-stationary behavior of the users.

 Banks usually maintain 40% excess of cash to evade cash out situations, but stocking of cash leads to high operational costs and interest rates. Artificial neural networks and support vector regression techniques were used to analyze the cash flow in atms and to reduce the amount of stocked cash to a minimal safe value.

The survey encompasses analysis and comparison of these two financial modelling techniques and recommends an effectual approach to optimize cash management in atms.