A web app that utilizes machine learning in order to detect exoplanet presence based on Kepler Objects of Interest Data (KOI) using several characteristics including orbital period, equilibrium temperature, planetary radius, and more.
Through KOI Data, this web app uses a DecisionTreeClassifier in order to predict whether or not a stellar object is indeed an exoplanet or a false positive. It uses several libraries such as scikit-learn, pandas, and matplotlib in order to clean, evaluate, and utilize data in order to create a well trained model.
In exoplanet detection, false positives can lead to misunderstandings and conclusions that may later affect future research. Due to this, there could be a waste of time and resources while trying to confirm false positives through extra observations. Machine learning can allow for the confirmation of false positives and also mitigating the effects by reducing usage of time and resources, making it a more efficient way of confirming false positives.
- Model trained on cleaned and preprocessed data, allowing maximum efficiency
- Usage of
joblibin order to save and load trained model - Evaluated using classification_report, accuracy_score, confusion_matrix, and roc_curve
- Uses a user-friendly user interface using
HTMLandCSS
git clone https://github.com/CookieCodesAI/Exoplant-Detection
cd Exoplanet-Detection
python3 -m venv venv
source venv/bin/activate #For Windows use venv\Scripts\activate
pip install -r requirements.txt
python app.py
This project uses data from the NASA Exoplanet Archive - Kepler Objects of Interest (KOI)
Source: NASA Exoplanet Archive - Kepler Objects of Interest:
https://exoplanetarchive.ipac.caltech.edu/cgi-bin/TblView/nph-tblView?app=ExoTbls&config=cumulative