Development of a warranty management system model using data-driven strategies for the Malaysian automotive industry

The increasing complexity of automotive systems, combined with large volumes of heterogeneous warranty data, presents significant challenges in achieving timely fault detection and effective warranty cost control. Conventional warranty management approaches remain predominantly reactive and are often constrained by data inconsistency, delayed reporting, and limited predictive capability. This study addresses these limitations through the development of a data-driven warranty management system model tailored to the Malaysian automotive industry, with a focused case study on Alternator Assembly components supplied by local company (Company MNC) to local car manufacturing company (Company A). A total of eight years of real-world warranty data were systematically managed through structured extraction, cleansing, validation, and classification processes to ensure data reliability and analytical consistency. Particular attention was given to resolving tangible data issues, including missing entries, duplicate claims, inconsistent failure coding, and non-trouble-found (NTF) cases arising from misdiagnosis. These corrections were essential to minimise bias and improve the robustness of subsequent analyses. Advanced analytical techniques, including statistical analysis, machine learning models, and data mining methods, were applied to identify failure patterns, temporal trends, and anomaly behaviours. Predictive modelling was validated using historical datasets, achieving acceptable performance with coefficient of determination (R² ≥ 0.60), demonstrating the capability of the proposed approach to handle real-world, imbalanced warranty data. However, the study also recognises inherent uncertainties influenced by external factors such as variations in service centre diagnostic practices, environmental operating conditions, and customer usage behaviour, which may affect model generalisation. In parallel, root cause analysis and component repair feasibility studies were conducted to evaluate cost-effective alternatives to full component replacement, supporting more targeted and sustainable warranty strategies. These analytical outputs were integrated into the Automotive Warranty Analysis System (AWAS), an interactive decisionsupport platform developed using Microsoft Power BI, enabling enhanced data transparency, early failure detection, and real-time monitoring. The results demonstrate measurable improvements in warranty management performance, including enhanced identification of high-risk components, reduction in unnecessary replacements through targeted repair strategies, and improved decision-making efficiency. The proposed framework also exhibits scalability and adaptability for broader industrial applications. This study contributes to the advancement of automotive warranty management by integrating predictive analytics, data governance, and engineering decision-making into a unified system. The findings support a transition from reactive to predictive warranty strategies, ultimately improving product reliability, reducing operational costs, and enhancing customer satisfaction.