Anonymous OEM achieves 80% accuracy predicting first-tier supplier disruptions with machine learning

The anonymous OEM, a complex asset manufacturer, set out to predict first-tier supply chain disruptions using the historical data available to it.

Researchers from the University of Cambridge, working with the anonymous OEM, developed a three-phase machine learning methodology applied to the company's historical data. The first phase focused on exploratory feature engineering — evaluating which attributes could serve as leading indicators of disruption. The engineered features included agility. The second phase developed a performance metric aligned with the specific goals of the case study. The third phase ran a systematic experimental design across multiple ML algorithm families and hyperparameter configurations, ensuring model selection was evidence-driven. No specific commercial vendor was identified; the methodology was developed as an academic-industrial collaboration.

The best-performing model achieved 80% accuracy in predicting late orders across first-tier suppliers. Adding the engineered agility feature outperformed the other experiments and led to the final algorithm. Key outcomes:

• 80% late-order prediction accuracy on the final algorithm
• Agility feature outperformed the other experiments, highlighting the importance of domain knowledge in feature construction

• Domain expertise is a first-class input: adding an engineered agility feature outperformed the other experiments — supply chain knowledge should inform feature design from the outset.
• Metric alignment matters: a performance metric aligned with the specific goals of the case study was used to rate the candidate methods.
• Systematic experimentation: testing different algorithms and parameters across the selected feature space led to the final algorithm.
• Historical data has predictive value: the OEM used historical data already available to it to predict late orders.