Forecasting solar energetic particles using multi-source data from solar flares, CMEs, and radio bursts with machine learning approaches

• DOI: 10.1038/s41598-025-92207-1
• Publication Date: 2025-03-19T19:17:17Z
• Authors: Mohammed AbuBakr Ali; Ali G. A. Abdelkawy; Abdelrazek M. K. ...; M. M. Beheary
• Keywords: Solar flare; Solar radio

This study presents a consistent method to the inherently imbalanced problem of predicting solar energetic particle (SEP) events, using variety datasets that include flares, coronal mass ejections (CMEs), and radio bursts. We applied several machine learning (ML) methods, including Random Forests (RF), Decision Trees (dtree), Support Vector Machines (SVM) with both linear (linSVM) nonlinear (svm) kernels. To assess model performance, we used standard metrics such as Probability Detection (POD), False Alarm Rate (FAR), True Skill Statistic (TSS), Heidke Score (HSS). Our results show RF consistently outperforms other algorithms across containing CMEs, For sweep frequency dataset, achieved POD [Formula: see text], FAR TSS text],and HSS text]). fixed-frequency produced text] ,and Key features for SEP prediction CME speed angular width datasets. frequency, flare intensity integral soft X-ray (SXR) flux are crucial, while fixed rise time duration bursts at 1415 MHz significant.