| # Predictive Maintenance – Tuned Random Forest Model | |
| ## Model Description | |
| This model is a tuned Random Forest classifier trained to predict engine maintenance requirements using sensor data such as RPM, oil pressure, fuel pressure, and temperature readings. | |
| ## Training Data | |
| - Dataset: Predictive Maintenance Engine Sensor Dataset | |
| - Source: Hugging Face Dataset Hub (`manjuprasads/predictive-maintenance-engine-data`) | |
| - Target Variable: `engine_condition` (0 = Normal, 1 = Maintenance Required) | |
| ## Model Objective | |
| The model prioritizes recall for engines requiring maintenance to minimize the risk of missed failures in safety-critical environments. | |
| ## Intended Use | |
| - Early detection of engine maintenance needs | |
| - Integration into real-time monitoring and alerting systems | |
| ## Limitations | |
| - The model is trained on snapshot sensor data and does not capture temporal trends. | |
| - Performance may vary across unseen engine types or operating regimes. | |
| ## Framework | |
| - scikit-learn | |
| ## Automated ML Pipeline | |
| This repository includes an automated machine learning pipeline that supports: | |
| - Data ingestion from Hugging Face dataset space | |
| - Preprocessing and feature preparation | |
| - Model training and evaluation | |
| - Model artifact registration | |
| The pipeline is implemented in a modular manner and is automation-ready. | |
| It can be triggered via CI/CD workflows (e.g., GitHub Actions) based on code or data changes. |