You can visit the live website here!
Here is the demo youtube video of the working site!
Team members: Supan Shah, Rudri Jani, Riya Shah, Dhaval Deshkar
The project trying to identify whether the given planet is livable planet or not using classification algorithms.
The data is collected from NASA's live API.
The calling of API can be found from the code.
The final dataset on which the model is being trained can be found in the /Dataset folder.
The public URL of the code can be found here
First the data is extracted using NASA's API. After some EDA and data cleaning and preprocessing, the feature selection is done. Using those features three classification models are implemented. Logistic Regression, Support Vector Machine and XGBoost is used to classify whether the planet is an exoplanet or not.
The final selected model is XGBoost using 5 features from NASA API viz. Disposition Score, TCE Planet Number, Planet Radius, Stellar Temperature and Transit signal-to-noise. The decision tree diagram looks like:
The images of the final results in our experiments is available in the /Results folder.
The trained model is deployed on the IBM cloud. It returns the probability of a planet being an exoplanet.
A user-friendly prediction website is deployed on heroku and the link is at the top of this readme.
