A predictive maintenance platform that uses telemetry sensor data and machine learning to identify server failure risk, visualize infrastructure health, and support maintenance decision-making.
- Random Forest failure prediction model
- SMOTE class balancing
- Interactive risk prediction dashboard
- Feature importance analysis
- Maintenance recommendation engine
- Failure risk gauge visualization
- ROC-AUC: 0.851
- Dataset Size: 124,494 telemetry records
- Failure Events: 106
- Class Imbalance Mitigated Using SMOTE
- Python
- Streamlit
- Scikit-Learn
- Pandas
- Plotly
Dataset not included due to size.
pip install -r requirements.txt
streamlit run app.py