Predicting material backorders in inventory management using machine learning

Material backorder is a common supply chain problem, impacting an inventory system service level and effectiveness. Identifying parts with the highest chances of shortage prior its occurrence can present a high opportunity to improve an overall company's performance. In this paper, machine learning classifiers are investigated in order to propose a predictive model for this imbalanced class problem, where the relative frequency of items that goes into backorder is rare when compared to items that do not. Specific metrics such as area under the Receiver Operator Characteristic and precision-recall curves, sampling techniques and ensemble learning are employed in this particular task.