Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Solar Flare Forecast: A Comparative Analysis of Machine Learning Algorithms for Solar Flare Class Prediction

Julia Bringewald

TL;DR

Solar flares pose significant space weather risks, motivating accurate prediction of GOES flare classes. The paper compares three classifiers—Random Forest, KNN, and XGBoost—on SHARP-derived magnetic parameters under two PCA regimes that capture 95% and 97.5% of the variance, evaluating both binary and multiclass GOES-class predictions. RF and XGBoost consistently deliver strong performance and gain from higher dimensionality, while KNN shows more limited improvements, especially in binary tasks. The findings inform dimensionality reduction strategies and model selection for astrophysical forecasting, with implications for real-time space-weather forecasting and solar-physics research.

Abstract

Solar flares are among the most powerful and dynamic events in the solar system, resulting from the sudden release of magnetic energy stored in the Sun's atmosphere. These energetic bursts of electromagnetic radiation can release up to 10^32 erg of energy, impacting space weather and posing risks to technological infrastructure and therefore require accurate forecasting of solar flare occurrences and intensities. This study evaluates the predictive performance of three machine learning algorithms: Random Forest, k-Nearest Neighbors (KNN), and Extreme Gradient Boosting (XGBoost) for classifying solar flares into 4 categories (B, C, M, X). Using the dataset of 13 SHARP parameters, the effectiveness of the models was evaluated in binary and multiclass classification tasks. The analysis utilized 8 principal components (PC), capturing 95% of data variance, and 100 PCs, capturing 97.5% of variance. Our approach uniquely combines binary and multiclass classification with different levels of dimensionality reduction, an innovative methodology not previously explored in the context of solar flare prediction. Employing a 10-fold stratified cross-validation and grid search for hyperparameter tuning ensured robust model evaluation. Our findings indicate that Random Forest and XGBoost consistently demonstrate strong performance across all metrics, benefiting significantly from increased dimensionality. The insights of this study enhance future research by optimizing dimensionality reduction techniques and informing model selection for astrophysical tasks. By integrating this newly acquired knowledge into future research, more accurate space weather forecasting systems can be developed, along with a deeper understanding of solar physics.

Solar Flare Forecast: A Comparative Analysis of Machine Learning Algorithms for Solar Flare Class Prediction

TL;DR

Solar flares pose significant space weather risks, motivating accurate prediction of GOES flare classes. The paper compares three classifiers—Random Forest, KNN, and XGBoost—on SHARP-derived magnetic parameters under two PCA regimes that capture 95% and 97.5% of the variance, evaluating both binary and multiclass GOES-class predictions. RF and XGBoost consistently deliver strong performance and gain from higher dimensionality, while KNN shows more limited improvements, especially in binary tasks. The findings inform dimensionality reduction strategies and model selection for astrophysical forecasting, with implications for real-time space-weather forecasting and solar-physics research.

Abstract

Solar flares are among the most powerful and dynamic events in the solar system, resulting from the sudden release of magnetic energy stored in the Sun's atmosphere. These energetic bursts of electromagnetic radiation can release up to 10^32 erg of energy, impacting space weather and posing risks to technological infrastructure and therefore require accurate forecasting of solar flare occurrences and intensities. This study evaluates the predictive performance of three machine learning algorithms: Random Forest, k-Nearest Neighbors (KNN), and Extreme Gradient Boosting (XGBoost) for classifying solar flares into 4 categories (B, C, M, X). Using the dataset of 13 SHARP parameters, the effectiveness of the models was evaluated in binary and multiclass classification tasks. The analysis utilized 8 principal components (PC), capturing 95% of data variance, and 100 PCs, capturing 97.5% of variance. Our approach uniquely combines binary and multiclass classification with different levels of dimensionality reduction, an innovative methodology not previously explored in the context of solar flare prediction. Employing a 10-fold stratified cross-validation and grid search for hyperparameter tuning ensured robust model evaluation. Our findings indicate that Random Forest and XGBoost consistently demonstrate strong performance across all metrics, benefiting significantly from increased dimensionality. The insights of this study enhance future research by optimizing dimensionality reduction techniques and informing model selection for astrophysical tasks. By integrating this newly acquired knowledge into future research, more accurate space weather forecasting systems can be developed, along with a deeper understanding of solar physics.
Paper Structure (17 sections, 9 figures, 4 tables)

This paper contains 17 sections, 9 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Dataset
  3. Methodology
  4. Experimental Setup
  5. Feature engineering
  6. Algorithms
  7. K-Nearest Neighbours (KNN)
  8. Random Forest classifier
  9. Extreme Gradient Boosting (XGBoost)
  10. Training and Evaluation
  11. Hyperparameters
  12. Notes on Hyperparameters per model:
  13. Results
  14. Model Performance with varying Principal Components
  15. Model Performance with Varying Classification Tasks (PC=8)
  16. ...and 2 more sections

Figures (9)

  • Figure 1: Covariance heatmap of magnetic parameters. Note: A covariance heatmap is a visual representation of the covariance matrix of a dataset's features. Covariance measures the strength of joint variability between two or more variables, as to say how much two variables change together. Positive covariance indicated that as one variable increases, the other tends to increase as well. Negative covariance indicates that as one variable increases, the other tends to decrease.
  • Figure 2: Cumulative Explained Variance by Principal Components (Interaction features)
  • Figure 3: Visualization of the feature engineering process
  • Figure 4: Visualization of models' training an evaluation method
  • Figure 5: Performance Curves KNN
  • ...and 4 more figures