Time Series Analysis for Prediction of Geomagnetic Indices

Published in Ludwig-Maximilians-Universität München, 2025

This study evaluates the three modelling approaches Cumulative Odds Model, XGBoost and LSTM to forecast the global geomagnetic index Hp30 up to 9 hours ahead, by applying a direct forecasing approach, which is provided by the forecastML R package. Therefore recently observed seasonal variations, solar wind data and Hp30 measurements are used as input data.