Design principles for the application of machine learning in supply chain risk management: an action design research approach

The opportunity to anticipate delivery failures, shortages or delays in a company’s upstream supply chains at an early stage facilitates to take preventive countermeasures to mitigate potential damage. However, data-driven predictive technologies such as machine learning (ML) are rarely examined in supply chain risk management (SCRM). The purpose of this paper is to present a framework of design principles for the application of ML in SCRM. The foundation of this framework is an action design research (ADR) project performed in collaboration with the SCRM department of an automotive company. A predictive ML model is developed and evaluated in collaboration with the company. Based on the findings and observations made during the project, general design principles are derived and grouped by the three interrelated elements of organization, development and operation, which are to be considered when applying ML in SCRM. Finally, the derived elements and the corresponding design principles are discussed and justified with reference to the literature.