Detecting and Classifying Solar Flares in High-Resolution Solar Spectra using Supervised Machine Learning

This research developed a full data pipeline for classifying solar flares, including processing high-resolution HARPS-N spectra, implementing Principal Component Analysis (PCA) for dimensionality reduction, and using undersampling to correct for data imbalance. We trained and optimized a C-Support Vector Classification (SVC) model with an RBF kernel, performing a GridSearch for hyperparameter tuning and validating the results with confusion matrices, precision, and recall.

Recommended citation: Nicole Hao, Laura Flagg, Ray Jayawardhana. (2024). "Detecting and Classifying Solar Flares in High-Resolution Solar Spectra using Supervised Machine Learning." The Astrophysical Journal.
Download Paper