---
title: Engine Predictive Maintenance App
emoji: "🛠️"
colorFrom: purple
colorTo: pink
sdk: docker
pinned: false
---

# 🛠️ Smart Engine Predictive Maintenance App

This interactive Streamlit application predicts whether an engine is likely to be **Faulty (1)** or **Normal (0)** using real-time sensor readings.  
It is designed to support **preventive maintenance decision-making** by identifying engines at higher risk of failure before breakdown occurs.

---

## ✅ Key Features

- **Single Engine Prediction** using manual sensor inputs  
- **Probability-based output** for Faulty / Normal (where supported by the model)  
- **Feature engineering built-in** (the app automatically computes engineered features to match the training schema)  
- **Download engineered input row** as CSV for traceability  
- **Bulk CSV Prediction** (upload a CSV and generate batch predictions)  
- **Download bulk predictions** directly from the UI

---

## 🧠 Model Details

- **Algorithm:** Gradient Boosting Classifier  
- **Training Data:** Engine sensor telemetry dataset  
- **Target Variable:** `Engine Condition`  
  - `0 = Normal`
  - `1 = Faulty`

**Reference Metrics (from model evaluation):**
- Recall (Faulty): ~0.84  
- ROC-AUC: ~0.70  
- PR-AUC: ~0.80  

---

## 🧾 Required Input Features (Single & Bulk)

Your CSV or manual inputs must include **only the raw sensor columns** below:

1. `Engine rpm`  
2. `Lub oil pressure`  
3. `Fuel pressure`  
4. `Coolant pressure`  
5. `lub oil temp`  
6. `Coolant temp`

The app computes additional engineered features internally (ratios, indices, and warning flags) to align with the model training pipeline.

---

## 📦 Bulk Prediction Instructions

1. Upload a CSV file with the 6 required raw sensor columns listed above.  
2. The app will generate:
   - `Predicted_Class` (0/1)
   - `Faulty_Probability` (if available)

3. Download the results using the provided **Download Bulk Predictions CSV** button.

---

## 🚀 Deployment

This Space uses a Docker-based deployment with Streamlit running on port **8501**. Hugging Face automatically maps ports during deployment.

---

## 🔗 Project Links

- **Model Hub:** `simnid/predictive-maintenance-model`  
- **Dataset Hub:** `simnid/predictive-engine-maintenance-dataset`  
- **GitHub Repository:** *(add your repo link here once finalized)*

---