Development of systems for the automatic analysis of fulfillment centers metrics

This thesis work deals with the development of two Machine Learning (ML) based systems for the automatic analysis of data in Amazon Fulfillment Centers (FCs), namely Amazon warehouses. The first system is dedicated to the forecasting of the amount of working hours needed to administer a shipment received in or to be sent from a FC. The goal of the project is to implement a tool to support FCs managers scheduling the work of Amazon associates and other resources. The second setup focuses on the identification of potential root causes affecting a FC performance. Starting from a dataset of defects registered in a FC, a clustering algorithm should detect anomalous quantities of errors sharing some common features. The clusters having the highest impact on the FC performance are reported to the managers, to allow them to trace the root causes behind the origin of the detected unusual amount of problems