- Machine Learning
- Strategy
Throughput is a critical performance metric for warehouse operations in the food industry. Accurate throughput estimations are necessary for effectively planning replenishments, inventory levels, and labor resources to meet the needs of customers. General Mills, who manages a large variety portfolio with different packaging, demand volatility, storage requirements, and outbound weight requirements, is interested in throughput estimation at their existing warehouses, also called Customer Service Facilities (CSFs). This project utilizes data collected from various data sources at General Mills to understand the factors that influence throughput. After interviewing key company stakeholders to learn more about warehouse operations, we collected and analyzed data. We developed a linear regression model, using machine learning to predict throughput. Ultimately, the analysis demonstrated that warehouse throughput at General Mills is not only impacted by internal factors, such as labor and product mix, but it is also impacted by external factors, such as day of the week, and higher demand requirements near quarter-end. With less than a year of data, the model still achieved a low mean absolute percentage error (MAPE) around 10%, implying highly accurate results. The strong forecast accuracy allows General Mills to create strategic plans to manage their labor constraints and improve the predictive performance of their throughput estimations.