Contextual Optimization of ATM Inventory

Research-to-Practice (with Siemens AG)

The optimization of ATM inventories can be tricky due to the added security and insurance issues when transporting money to refill ATMs. In a joint prescriptive and predictive approach (contextual optimization), we aim to explore factors that help us to make better decisions regarding the optimal inventory of ATMs (inventory management) and their replenishment (vehicle routing) by building a predict-then-optimize approach. We test multiple features and machine learning models with the goal to improve decision performance rather than pure predictive performance of the demand forecast.

Partner: Siemens AG