Forecasting the State of the Earth’s Magnetosphere Using a Special Algorithm for Working with Multidimensional Time Series

This study is devoted to the adaptation and application of a special multistage algorithm based on machine learning methods, developed for the analysis of multidimensional time series in solving problems of forecasting certain events and identifying their precursors—phenomena represented by an unknown combination of parameter values describing an object. In addition to forecasting events, the algorithm can be used to forecast the values of continuous quantities. In this study, we compare the results of application of this algorithm in forecasting of three physical quantities characterizing the state of the magnetosphere of the Earth—two geomagnetic indexes (Dst and Kp) and the flux of relativistic electrons ( $$E>$$ 2 MeV) in geostationary orbit.