You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
End-to-end time series analysis and forecasting of monthly sunspot numbers using classical machine learning and deep learning models, including linear regression, tree-based methods, and advanced PyTorch architectures.
This project predicts sunspot activity using an LSTM model for time series data. Built with TensorFlow and Keras, it uses Huber loss for outlier handling and MAE for performance evaluation. The dataset, sourced from Kaggle or SIDC, spans over 270 years of monthly sunspot data.
Project aims at modeling the size distribution of sunspots greater than 60 millionths of a solar hemisphere (MSH) using a truncated log-normal distribution.