From industrial data to business decisions: Industrial case studies as teaching instruments in interdisciplinary analytics education

This paper analyses industrial problems as teaching instruments in interdisciplinary analytics education, based on 14 student project reports from two industrial case studies: SICK Mobilisis (AI-based fuel-event detection from vehicle telemetry data) and Emil Frey Digital (digital vehicle sales optimisation). Reports were analysed using a qualitative coding framework covering business-process understanding, analytical-unit construction, target design, feature availability, validation, metric interpretation, decision support, implementation, and risk. Students usually completed expected sections but had more difficulty connecting them into a coherent process from dana to decision, particularly regarding analytical-unit definition, leakage-aware feature selection, validation design, and linking model outputs to business decisions.