Skip to content

A machine learning project that detects potential exoplanets using a Random Forest model trained on NASA’s Kepler Object of Interest (KOI) dataset. It analyzes stellar and orbital parameters to classify confirmed exoplanets vs. false positives with high accuracy.

License

Notifications You must be signed in to change notification settings

ashwinraj8090/exoplanet_koi

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CYGNUS

An Exoplanet Detector

A machine learning project built for the NASA Space Apps Challenge, designed to detect potential exoplanets using a Random Forest classifier trained on NASA’s Kepler Object of Interest (KOI) dataset.

The dataset contains nearly 10,000 celestial objects, each described by 17 physical and orbital parameters. The model uses these features to predict whether a given object is likely an exoplanet or a false positive, inspired by the classic transit detection method used in astronomy.


Live Demo

Try it out here 👉 Exoplanet Detector Website

You can input the 17 parameters directly on the website and instantly get a prediction on whether the object is an exoplanet or not.

⚠️ Note: The live demo may be taken down in the future as server availability may change.


Model Overview

  • Algorithm: Random Forest Classifier
  • Dataset: NASA Kepler Object of Interest (KOI) dataset
  • Parameters: 17 stellar/orbital parameters per object
  • Goal: Classify whether a celestial body is a confirmed exoplanet
  • Model Accuracy: 93%

The project includes complete preprocessing, training, and evaluation code, along with a ready-to-use model.


⚙️ How to Run Locally

1️⃣ Clone the repository

git clone https://github.com/arungjose/exoplanet_koi.git
cd exoplanet_koi

2️⃣ Create and activate a virtual environment

Windows:

python -m venv venv
venv\Scripts\activate

macOS / Linux:

python3 -m venv venv
source venv/bin/activate

3️⃣ Install dependencies

pip install -r requirements.txt

4️⃣ Run the demo script

python test_model.py

This script runs the pre-trained model with a predefined set of parameters for an exoplanet (you can modify them in the script to test your own data).


🧪 Method Used – The Transit Method

The project is inspired by the transit photometry technique, where periodic dips in a star’s brightness indicate a planet passing in front of it. The AI model learns from these patterns in the Kepler dataset to automate and refine detection accuracy.


💡 Contribution

Feel free to clone, experiment, and develop this project further.
If you improve the model, visualization, or UI — contributions are always welcome.

Happy Coding ❤️

About

A machine learning project that detects potential exoplanets using a Random Forest model trained on NASA’s Kepler Object of Interest (KOI) dataset. It analyzes stellar and orbital parameters to classify confirmed exoplanets vs. false positives with high accuracy.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • HTML 71.3%
  • Python 28.0%
  • Dockerfile 0.7%