Business rule extraction using decision tree machine learning techniques: A case study into smart returnable transport items

Decision support systems are becoming increasingly sophisticated (e.g., being machine learning-based), attempting to automate decisions as much possible. However, it remains challenging extract meaningful value from large quantities of data while also maintaining transparency in seeking justification for the choices made. Instead creating methods increasing interpretability black box models, one way forward is design models that inherently interpretable first place. Rule-based can with great and accuracy, helping ensure compliance regulations adherence organizational guidelines. In this paper, we propose an approach uses a decision tree learning classification technique extracting business rules IoT-generated predict asset status Smart Returnable Transport Items (SRTIs). We report on industrial case study two years historical data, obtained SRTI provider Netherlands, smart pallets. compare performance results by using support-vector (SVM) technique. Our experiments show our solution both accurate flexible terms rule elicitation. The trees human-interpretable, easily be combined other decision-making techniques, provide prediction accuracy marginally higher than SVM