ATM Chest Cash Demand Prediction Using Enhanced HistGB Regression Model

Abstract

Automated teller machine, ATM, an efficient fintech invention in catering customer to withdraw money, by not visiting branch. ATM services are expanded to all bank customers and there by each financial institution, in addressing customer service, faces couple of challenges ie (i) availability of cash in the machine for customer access 24X7, as banks cannot load inappropriate cash levels, because of bin size as well as cash usage pattern. Less cash level hit on the customer service and more level hits on investment, thereby revenue earning to bank. Hence, individual ATM cash demand prediction is turned out to be most essential to financial institutions. (ii) tasks of Cash replenishment in ATMs. The cash loading or cash replenishment services are handled by bank staff on demand basis. However, the growth of off-site ATMs, ie the installation of ATMs on the sites of most people gather like Malls, prominent theatres, petrol bunks etc. where in physical security is in built, force the banks to outsource this activity to third party to provide continued customer service by ATMs. Hence it is important to Banks, through any means, to predict the volume of cash requirement, in advance, to replenish in the set of ATMs, so that with minimum transportation efforts, cash filling can be done on the set of ATMs. Information technology comes handy to this prediction. Software based algorithms are more configurable than the statistical application of limited approach of earlier years. This paper explains that the enhanced HistGB regressor algorithm is found out to be suitable for the prediction of volume of cash needs to replenish for set of ATMs to be helpful to banks cash chest or outsourcing contract.