MACHINE LEARNING APPLICATIONS TO MINIMIZE DELIVERY DAMAGES IN SUPPLY CHAIN OPERATIONS

Abstract

The e-commerce and supply chain industries face significant challenges in ensuring undamaged deliveries to customers. Traditional approaches, which rely on experience and historical data, often fail to capture the latest trends and complex interactions among factors contributing to damaged shipments. This paper proposes a machine learning-based solution for improved detection and prevention of damaged deliveries. This paper presents the development of multiple classification models that predict the probability of damages using a comprehensive set of features. The models are trained on the latest data to capture trends across all US locations. The proposed approach emphasizes model interpretability and stakeholder buy-in through effective communication of business insights and model explainability. The implementation of the machine learning solution can lead to several performance improvement projects, such as switching carriers, adding new delivery hubs, and negotiating e-commerce-ready packaging with vendors. Future work includes continuous monitoring and improvement of ML operations, as well as further investigation of root causes and new projects for sustained performance enhancements. By leveraging machine learning, the proposed approach offers a predictive and proactive solution to the pervasive issue of damaged deliveries in e-commerce and supply chain operations.