In recent years, the application of machine learning techniques in the field of predictive vehicle maintenance has emerged as a highly effective approach for reducing unexpected breakdowns, enhancing operational safety, and optimizing maintenance and repair costs. Predictive maintenance aims to anticipate potential failures before they occur, thereby minimizing vehicle downtime, reducing maintenance expenses, and improving overall system reliability. The primary objective of this study is to develop, implement, and evaluate a range of machine learning models for predicting vehicle maintenance needs using operational and performance data collected from various vehicle subsystems, including engine, transmission, braking, and electrical systems.



The dataset used in this research underwent thorough preprocessing, including data cleaning, normalization, and handling of missing values, followed by a division into 80% training and 20% testing sets. Twelve machine learning algorithms were implemented and compared, comprising traditional methods such as Support Vector Machine (SVM), Logistic Regression (LR), Naïve Bayes (NB), k-Nearest Neighbors (KNN), Decision Tree (DT), and Random Forest (RF), as well as ensemble methods including AdaBoost, Bagging, Stacking, XGBoost, CatBoost, and LightGBM. The models’ performance was systematically evaluated using multiple metrics: accuracy, precision, recall, F1-score, and ROC-AUC, providing a comprehensive assessment of both classification ability and predictive reliability.



The experimental results demonstrated that tree-based algorithms and ensemble approaches consistently outperformed other methods in terms of predictive accuracy and robustness. Among them, LightGBM achieved the highest performance, with an AUC of 0.9475 and an F1-score of 0.9613, indicating superior capability in capturing complex patterns and correlations in vehicle operational data. The findings of this research highlight the significant potential of machine learning techniques—particularly ensemble models—in predictive maintenance applications. These methods can serve as a foundation for designing intelligent maintenance systems that enhance vehicle reliability, reduce operational costs, and support proactive decision-making in automotive fleet management.