# 🚀 NASA Predictive Maintenance

Predicting engine failure before it happens.  
Using sensor data from NASA’s CMAPSS dataset to estimate Remaining Useful Life (RUL) of aircraft engines.

---

## 📊 About the Project

- ⏱ Predict how many cycles remain before failure (RUL)
- 🔧 Enables smarter maintenance scheduling
- 📉 Reduces unplanned downtime & costs

---

## 🧠 Tech Stack

- Python (Pandas, NumPy)
- XGBoost Regressor
- Matplotlib, Seaborn
- Jupyter Notebook

---

## 🔁 Workflow

1. **Data Loading** – NASA turbofan dataset  
2. **Preprocessing** – Cleaned, normalized, created RUL  
3. **EDA** – Explored patterns & trends  
4. **Modeling** – XGBoost trained on sensor data  
5. **Evaluation** – RMSE & R² Score  
6. **Feature Importance** – Identified key sensors

---

## 🚦 Results

- 📈 R² Score: ~0.91  
- 🎯 RMSE: ~15 cycles  
- 🔍 Top Features: Sensor 15, Sensor 4, Sensor 11

---

## ⚙️ Use Case

Real-time predictive maintenance for aerospace, automotive, or heavy machinery industries.  
From failure-prone → failure-proof.

---

## 🙋‍♂️ Author

**Rohit Mannur**  
🔗 [LinkedIn](https://linkedin.com/in/rohit-mannur-851a82288) • 🐙 [GitHub](https://github.com/rohitmannur007)

---

> "Predict before it breaks. Maintain before it fails."