Demand Forecasting in the Food Industry: a Real Case Study

The present project was elaborated during an internship in the company SDG Consulting Italia S.p.A. motivated by the need of a client company, leader in the national food market, to enhance the demand forecasting approach currently in place. The thesis focuses on the description of the statistical methods employed, particularly focusing on the distinction between baseline and uplift prediction, and on the way they were implemented and integrated within the company's systems. A key topic is how the forecasting algorithm's results are integrated with planners' domain knowledge and made accessible for review and adjustments via the Board platform. Starting from here, the thesis discusses further developments aimed at improving forecasting accuracy of results by increasing the flexibility in the approach, better addressing the high degree of variability intrinsic to the food industry, both in terms of distribution channel and product type.