CAOSP abstracts, Volume: 53, No.: 3, year: 2023

Abstract: In presented work we further explore previously indicated possibility of the existence of two classes of Forbush decrease events, established by the prior analysis of the correlation between the shape of energetic proton fluence spectra and Forbush decrease properties. In an attempt to increase statistical robustness of the analysis and potentially reduce the uncertainties, we have developed an alternative classification procedure that employs machine learning and utilizes space weather parameters as input variables. Based on the overall performance, efficiency and flexibility of different machine learning methods we selected the best performing algorithm and established the optimal boundary value of Forbush decrease intensity to be used for class separation. A subset of good input variables was selected based on their predictive power.

Full text version of this article in PDF.


Back to:
CAOSP Vol. 53 No. 3 index
CAOSP archive main index
CAOSP main page
Astronomical Institute home page
Valid XHTML 1.0! Valid CSS!

Last update: November 30, 2023