Sells Optimization Through Product Rotation

Abstract

This article describes a real use cases of applications of Machine Learning (ML) and Optimization techniques in the business context.

Following the business requests, the Proofs of Concept were developed first and then put into production, rethinking the existing processes and integrating the role of these algorithms in the decision-making processes.

The use case goal is to maximize the overall revenues through the rotation of Vending Machines (VMs) within a network of points of sale (PoS), the VMs sell different products. A new product in a PoS brings a revenue increase in the first weeks. To estimate the expected revenue in the new PoS, in the first n weeks, for all combination of VMs exchange, two ML models have been used.

After the revenue’s calculation for all possible exchange, to choose the optimal transfer chain and respect the business constraints, an optimization model has been used. The optimization model has been formulated as a goods transportation problem, where each node (VM position) on the network can be a source or destination node.