A machine learning approach for predictive warehouse design

Abstract Warehouse management systems (WMS) track warehousing and picking operations, generating a huge volumes of data quantified in millions to billions records. Logistic operators incur significant costs maintain these IT systems, without actively mining the collected monitor their business processes, smooth flows, support strategic decisions. This study explores impact tracing beyond simple traceability purpose. We aim at supporting design system by training classifiers that can predict storage technology (ST), material handling (MHS), allocation strategy (SAS), policy (PP) system. introduce definition learning table, whose attributes are benchmarking metrics applicable any Then, we investigate how availability warehouse (i.e. varying number table) affects accuracy predictions. To validate approach, illustrate generalisable case which collects from sixteen different real companies belonging industrial sectors (automotive, manufacturing, food beverage, cosmetics publishing) players (distribution centres third-party logistic providers). The applied used generate tables with attributes. A bunch is identify crucial input prediction ST, MHS, SAS, PP. managerial relevance data-driven methodology for showcased 3PL providers experiencing fast rotation SKUs stored systems.