NASA RUL Prediction

Machine learning models for predicting Remaining Useful Life (RUL) on the NASA C-MAPSS turbofan degradation dataset.
Achieved RMSE: 2.79, demonstrating top-tier performance on this benchmark.

Made for a presentation to illustrate python and machine learning with test data.

https://data.nasa.gov/Aerospace/CMAPSS-Jet-Engine-Simulated-Data/ff5v-kuh6/about_data

---

---

✨ Features

• Preprocessing pipeline for time-series sensor data
• Feature engineering (statistical, temporal, and degradation trends)
• Implementation of regression models and ensembles
• Training/validation splits aligned with NASA data structure
• Performance metric: Root Mean Squared Error (RMSE)

---

🛠 Tools & Libraries

• Python 3.8
• pandas, numpy, scikit-learn
• matplotlib, plotly, seaborn (visualization)
• jupyter (tutorials & exploration)

---

📊 Results

• Final model: Ensemble regressor
• Test RMSE: 2.79
• Outperforms common baselines and demonstrates robustness across multiple engine units

---

Initial Baseline Run

• Baseline Run before feature engineering

Final Run with Ensemble Method

• Final run with ensemble method and feature engineering and RMSE of 2.79

Bonus: Plotly Plot inside PySide GUI

First time putting an interactive plot in a GUI. Have improved my methods greatly since.

Click to play 👇

---

🚀 How to Run

Clone the repo and install dependencies:

git clone https://github.com/ratrent55/NASA_RUL.git
conda env create -f environment.yml