Ensemble Learning for Solar Flare Intensity Prediction: Enhancing Accuracy and Efficiency with Xgboost and Gradient Boosting

Solar flares are significant eruptions of electromagnetic energy from the Sun that can disrupt terrestrial communication systems, satellites, and power grids. Accurate and timely prediction of solar flare intensity is crucial to mitigate their potential impact on critical infrastructures. This research introduces a machine learningbased framework utilizing ensemble learning methods, specifically Gradient Boosting and XGBoost algorithms, to predict the intensity of solar flares. By analyzing multivariate solar activity data, including features such as peak counts, positional coordinates, and flare duration, the system constructs robust predictive models. Preprocessing involves data cleaning and feature extraction to ensure model efficiency. Our study highlights that while traditional machine learning algorithms are accurate but time-intensive, and deep learning techniques often lack efficiency for this context, ensemble learning methods provide a superior balance of accuracy and computational efficiency. Among the tested methods, Gradient Boosting and XGBoost consistently delivered the best results. The prediction system incorporates a Flask-based frontend offering an interactive interface for real-time solar flare intensity predictions. Users can input solar activity data and receive predictions along with visualized results. This project's innovation lies in its ability to leverage advanced ensemble learning techniques for safeguarding essential systems against solar disturbances.