Systematic analysis of machine learning techniques for Kp prediction

The Kp index is a global measure of geomagnetic activity and it represents short-term magnetic variations driven by space weather. The Kp index is used as an input to various thermosphere and radiation belt models, and it is therefore important to predict it accurately. In this study, we systematically test how different machine learning techniques perform on the task of nowcasting and forecasting Kp for 3, 6, and 9 hours prediction horizons. Additionally, we investigate two feature selection schemes based on Mutual Information and Random Forests. Finally, we evaluate and report the optimal combinations of input parameters and the best performing machine learning model.