SMART ma’lumotlari asosida disk nosozliklarini mashina o‘rganish modellari yordamida bashoratlash
Keywords:
SMART monitoring, disk failure, predictive maintenance, machine learning, Random Forest, XGBoost, anomaly detection, LSTM, class imbalance, F1-scoreAbstract
This article evaluates machine-learning models for predicting disk failures from SMART monitoring data. Random Forest, XGBoost, SVM, Isolation Forest, Autoencoder, and LSTM were compared using a model dataset of 40,000 disk-day observations with imbalanced classes. The results indicate that Random Forest provides the most balanced F1 performance for static features, whereas LSTM has strong potential when temporal sequences are available. Model-selection conditions are identified, and a conceptual three-layer hybrid architecture is proposed for practical predictive maintenance.
References
Murray J.F., Hughes G.F., Kreutz-Delgado K. Machine Learning Methods for Predicting Failures in Hard Drives: A Multiple-Instance Application // Journal of Machine Learning Research. – 2005. – Vol. 6. – P. 783–816.
Tomic A., Pejic Bach M. Anomaly Detection Model for Predicting Hard Disk Drive Failures // Applied Artificial Intelligence. – 2021. – Vol. 35, No. 15. – P. 1153–1172.
Breiman L. Random Forests // Machine Learning. – 2001. – Vol. 45, No. 1. – P. 5–32.
Chen T., Guestrin C. XGBoost: A Scalable Tree Boosting System // Proceedings of the 22nd ACM SIGKDD Conference. – 2016. – P. 785–794.
Liu F.T., Ting K.M., Zhou Z.-H. Isolation Forest // Proceedings of the 8th IEEE International Conference on Data Mining. – 2008. – P. 413–422.
Hochreiter S., Schmidhuber J. Long Short-Term Memory // Neural Computation. – 1997. – Vol. 9, No. 8. – P. 1735–1780.
Cortes C., Vapnik V. Support-Vector Networks // Machine Learning. – 1995. – Vol. 20, No. 3. – P. 273–297.
Goodfellow I., Bengio Y., Courville A. Deep Learning. – Cambridge: MIT Press, 2016. – 800 p.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Shamurod Xolmurov Abdihamidovich
This work is licensed under a Creative Commons Attribution 4.0 International License.
