Estimation of Semimajor Axis of Exoplanet Orbit Using Machine Learning Techniques

Semimajor axis is one of the most important orbital parameters of an exoplanet. It is used to calculate planet signatures like insolation flux, which may inform us about the existence of life on exoplanets. The semimajor axis is calculated using Kepler's third law of motion and techniques like transits, radial velocity etc., which require the orbital period of the planet to be known. It is a time-intensive process to precisely observe long orbital periods. We aim to bridge this gap by building a machine learning model that estimates the semimajor axis using relevant features of the exoplanets which were explored in the past and archived as NASA Kepler mission data. So far we know, this initiative is the first of its kind. A Light Gradient Boosting based regression framework is developed that uses correlation maps to remove similar features to avoid overfitting. Furthermore, noisy features are removed and one-hot-encoding scheme is used for categorical features. The performance of the proposed regression framework is used to estimate the semimajor axis of the NASA Kepler mission data, whose performance is compared with other regression techniques. The empirical analysis shows that this framework outperforms the other techniques in terms of root mean squared error and R2 score. In the future, we aim to explore how this model behaves for specific types exoplanetary systems in order to find the exoplanet data category for which this model gives optimum results.