This repository is meant to act on the NASA Exoplanet Archive information in order to predict the log radius, in terms of Earth radii, of exoplanets by using and applying various Machine Learning techniques.
In order to run the code, preprocessing must first be done by running the Preprocessing.ipynb file - which also contains the baseline models and accuracy - as well as the file for regularization to generate the full datasets. From here RANDOM_FOREST.ipynb, PCA_Gradient_Boosting_Trees.ipynb, and Gradient_Boosting_Trees.ipynb are capable of being ran to fully to train and evaluate these models. Prior to the running of the Gradient_Boosting_Trees_Post_Lasso.ipynb and the RANDOM_FOREST_Feature_Engineering.ipynb, the file Linear_with_feature_engineering.ipynb. The Demo.ipynb can also be run to create the best model and evaluate its performance on data points.
The following packages must be installed in order for all relevant code to be fully runnable
numpy
pandas
sklearn
matplotlib
plotly
All data was sourced from https://exoplanetarchive.ipac.caltech.edu/cgi-bin/TblView/nph-tblView, and is present in ExoplanetArchiveAll.csv and ExoplanetArchiveValuesOnly.csv. All data resulting from this is processed internally to create multiple other .csv files that can be used in model training.
The data must be placed in the same folder as the scripts, as they expect all relevant csv's to share a base directory with them.