Exploring Exoplanets using kNN, Logistic Regression and Decision Trees

When compared to the vastness of space, our Solar System is of microscopic size. Millions of astrophysical objects exist outside of our Solar System and are being studied. Out of all those objects, studying Exoplanets is of utmost importance. An Exoplanet is a planet that exists out of our Solar System, which means they don’t revolve around our Sun. But it is not always easy to identify whether a given astrophysical object is an Exoplanet. Planets don’t emit light of their own, but the stars they orbit do. When these stars are observed for very long periods of time, we notice a periodic fluctuation of light intensity, also known as the flux of the star. This may be evidence that there may be Exoplanets orbiting the star. Exoplanet identification and confirmation is a task that is usually carried out exclusively by teams of astronomers and astrophysicists who have extensive education and training in their field. NASA’s Kepler Mission has revolutionized the field of Exoplanet Exploration and has enabled researchers to study extrasolar objects of interest. The data harnessed by these satellites are readily available for any individual to use. In our study we use several machine learning algorithms to analyse these data sets and train a model capable of accurately predicting whether an astrophysical body is an Exoplanet or not. The goal of our study is to compare the efficiencies of various machine learning algorithms with respect to detecting an exoplanet.