In a supply chain that produces thousands of products a day, with a huge network of processes to build up the final product, identifying the most common causes of delays on the product delivery can be very tricky. In a big company, saving even minutes per product could increase income by thousands of euro a day.
Extracting and understanding information to find the bottleneck within this kind of supply chain is a very complex task. Federico and his colleagues are trying to apply machine learning techniques with which they can approximate and simulate the production process in order to simplify the data analysis. Furthermore, machine learning techniques allow them to learn and try new policies for production by simulation, in order to see the corresponding effects before acting in the real world.