Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques
Abstract Predictive maintenance is an important application in the automotive industry to enhance vehicle reliability and reducing operational downtime. However, the major challenge with the predictive maintenance types of datasets is the class imbalance, where failure instances are scarce. In this...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-04-01
|
| Series: | Discover Applied Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s42452-025-06827-3 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849737562983235584 |
|---|---|
| author | Yashashree Mahale Shrikrishna Kolhar Anjali S. More |
| author_facet | Yashashree Mahale Shrikrishna Kolhar Anjali S. More |
| author_sort | Yashashree Mahale |
| collection | DOAJ |
| description | Abstract Predictive maintenance is an important application in the automotive industry to enhance vehicle reliability and reducing operational downtime. However, the major challenge with the predictive maintenance types of datasets is the class imbalance, where failure instances are scarce. In this study, a binary classification task is experimented leveraging advanced data imbalance handling techniques to predict the failure instances. The on-board diagnostic dataset utilized has only 16.3% of the failure data, and to address this, 3 key approaches were explored: [i] synthetic minority oversampling technique (SMOTE), [ii] cost-sensitive learning, [iii] ensemble methods. Six machine learning models, including logistic regression, support vector machine, decision tree, and random forest, along with gradient boosting algorithms using extreme gradient boost (XGBoost) and light gradient boosting machine frameworks, were implemented. SMOTE showed a statistically significant improvement in the F1-score with a significance value of 0.0010 and ROC-AUC with p = 0.0015, confirming the effectiveness of SMOTE in handling class imbalance. Advanced ensemble models like RUSBoost, balanced bagging, easy-ensemble, and balanced random forest were implemented where XGBoost yielded the highest performance, attaining scores of 1.000. for precision, recall and F1-score. SMOTE improved minority class representation, achieving a f1-score of up to 99.54%. The integration of multiple imbalance-handling techniques demonstrated superior performances. The scope of the study is limited to engine parameters; however, the results demonstrate the effectiveness of combining traditional and ensemble classifiers to handle imbalance. This work provides a reliable framework with potential applications in fleet management, automotive manufacturing, and smart vehicle diagnostics and also extends beyond engine performance monitoring to environmental and safety assessments. |
| format | Article |
| id | doaj-art-045a5040727e4768ac8b33307f31ac8d |
| institution | DOAJ |
| issn | 3004-9261 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Springer |
| record_format | Article |
| series | Discover Applied Sciences |
| spelling | doaj-art-045a5040727e4768ac8b33307f31ac8d2025-08-20T03:06:52ZengSpringerDiscover Applied Sciences3004-92612025-04-017412110.1007/s42452-025-06827-3Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniquesYashashree Mahale0Shrikrishna Kolhar1Anjali S. More2Symbiosis Institute of Technology, Symbiosis International (Deemed University)Symbiosis Institute of Technology, Symbiosis International (Deemed University)Symbiosis Institute of Technology, Symbiosis International (Deemed University)Abstract Predictive maintenance is an important application in the automotive industry to enhance vehicle reliability and reducing operational downtime. However, the major challenge with the predictive maintenance types of datasets is the class imbalance, where failure instances are scarce. In this study, a binary classification task is experimented leveraging advanced data imbalance handling techniques to predict the failure instances. The on-board diagnostic dataset utilized has only 16.3% of the failure data, and to address this, 3 key approaches were explored: [i] synthetic minority oversampling technique (SMOTE), [ii] cost-sensitive learning, [iii] ensemble methods. Six machine learning models, including logistic regression, support vector machine, decision tree, and random forest, along with gradient boosting algorithms using extreme gradient boost (XGBoost) and light gradient boosting machine frameworks, were implemented. SMOTE showed a statistically significant improvement in the F1-score with a significance value of 0.0010 and ROC-AUC with p = 0.0015, confirming the effectiveness of SMOTE in handling class imbalance. Advanced ensemble models like RUSBoost, balanced bagging, easy-ensemble, and balanced random forest were implemented where XGBoost yielded the highest performance, attaining scores of 1.000. for precision, recall and F1-score. SMOTE improved minority class representation, achieving a f1-score of up to 99.54%. The integration of multiple imbalance-handling techniques demonstrated superior performances. The scope of the study is limited to engine parameters; however, the results demonstrate the effectiveness of combining traditional and ensemble classifiers to handle imbalance. This work provides a reliable framework with potential applications in fleet management, automotive manufacturing, and smart vehicle diagnostics and also extends beyond engine performance monitoring to environmental and safety assessments.https://doi.org/10.1007/s42452-025-06827-3Artificial intelligenceBinary classificationClass imbalanceMachine learningPredictive maintenanceVehicle diagnostics |
| spellingShingle | Yashashree Mahale Shrikrishna Kolhar Anjali S. More Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques Discover Applied Sciences Artificial intelligence Binary classification Class imbalance Machine learning Predictive maintenance Vehicle diagnostics |
| title | Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques |
| title_full | Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques |
| title_fullStr | Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques |
| title_full_unstemmed | Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques |
| title_short | Enhancing predictive maintenance in automotive industry: addressing class imbalance using advanced machine learning techniques |
| title_sort | enhancing predictive maintenance in automotive industry addressing class imbalance using advanced machine learning techniques |
| topic | Artificial intelligence Binary classification Class imbalance Machine learning Predictive maintenance Vehicle diagnostics |
| url | https://doi.org/10.1007/s42452-025-06827-3 |
| work_keys_str_mv | AT yashashreemahale enhancingpredictivemaintenanceinautomotiveindustryaddressingclassimbalanceusingadvancedmachinelearningtechniques AT shrikrishnakolhar enhancingpredictivemaintenanceinautomotiveindustryaddressingclassimbalanceusingadvancedmachinelearningtechniques AT anjalismore enhancingpredictivemaintenanceinautomotiveindustryaddressingclassimbalanceusingadvancedmachinelearningtechniques |