ATM Cash Optimization Using Machine Learning

The Problem

Despite widespread perception that usage of cash will come to an end soon, the value and volume of cash continues to increase yearly across different countries. Banks spend a substantial amount of money for supporting cash circulation through the deposit of cash at counters and withdrawal of cash via ATMs.

There can be several thousand ATMs for a bank in a country. Keeping optimum levels of cash at each ATM is in the best interest of the bank. Keeping too much cash at an ATM means there is unused cash which is not generating revenues for the bank. Keeping low cash would result in cash out leading to poor customer service. So how do banks decide how much money to keep at ATMs daily?

Solution

Machine Learning models can be deployed on Kranium to optimizes cash usage across ATMs. Predictive models can be used to forecast cash usage. The model can be trained to take various parameters into consideration like Holidays, Weather, Historical Data, Special Events, Geo Location, Crisis Situations and more. In addition, machine learning models can also be deployed on Kranium to optimize the cost of transporting of cash.