Optimizing Recall Campaigns in the Automotive Industry: A Survival Analysis and Machine Learning Approach

This thesis explores the application of survival models and Machine Learning techniques to improve recall campaign management in the automotive sector. The adoption of the Accelerated Failure Time (AFT) model, enriched with technical and contextual variables, enabled accurate predictions of failure rates for vehicle components. This adaptable model, tailored to specific market characteristics and vehicle usage patterns, proved effective in optimizing recall costs by reducing waste and focusing interventions on high-risk vehicles. The integration of the model on serverless platforms allowed for near real-time data processing, ensuring operational flexibility and supporting dynamic decision making for resource management. The results demonstrated significant benefits, including reduced operational costs, improved recall campaign efficiency, and a proactive approach to preventive maintenance. This work provides a robust and scalable foundation for future applications, opening the door to real-time data based developments and deep learning techniques for further optimization of recall campaigns and product lifecycle management.