Enhancing supply chain security with automated machine learning

Abstract Increasing scale and complexity of global supply chains have led to new challenges spanning a variety of fields such as supply chain disruptions due to long waiting lines at the ports, material shortages, and inflation. Coupled with the size of supply chains and availability of vast amounts of data, efforts towards tackling such challenges have led to an increasing interest towards the application of machine learning methods in many aspects of supply chains. Compared to other solution methods, machine learning methods particularly perform better in making predictions based on a set of observations and approximating optimal solutions faster. Machine learning methods are called for analyzing very large datasets. This paper presents an automated machine learning framework to enhance the supply chain security such as detection of fraudulent activities, prediction of maintenance needs, and material backorder prediction. Results indicate that many factors affect the performance of ML methods such as sampling method, encoding categorical values, feature selection, hyperparameter optimization for different algorithms. In general, the number of variables poses a limit for mathematical programming models to performance on large-scale problem. The automated machine learning framework streamlines the processes including data processing, models construction, hyperparameter optimization and inference deployment. This paper contributes to the body of knowledge on supply chain security by developing an automated machine learning framework to detect fraud and to predict supply chain maintenance needs and material backorder.