Rashomon Perspective for Measuring Uncertainty in the Survival Predictive Maintenance Models

Predicting the Remaining Useful Life of aircraft engines is critical in high-reliability sectors such as aerospace and defense. Early failure predictions help reduce maintenance costs and prevent unexpected failures. Traditional regression models struggle with censored data, leading to biased predictions, whereas survival models effectively handle censoring, improving predictive accuracy. This paper introduces a novel approach based on the Rashomon perspective, which considers multiple models with similar performance rather than relying on a single best model. This enables uncertainty quantification in survival predictions and enhances decision-making in predictive maintenance. Results on the CMAPSS dataset show that depending solely on one model for Remaining Useful Life estimation may increase risk. Higher censoring levels amplify prediction uncertainty, with longer censoring times causing greater variability in survival probabilities. These findings highlight the importance of model multiplicity in predictive maintenance to achieve more robust failure predictions.