Machine Learning on Multivariate Time Series for Solar Flare Prediction

Junzhi Wen
Advisor: Dr. Rafal Angryk

In this project, we use an SVM classifier with different kernels on time-series data and extract statistical features from the SWAN-SF multivariate time-series benchmark dataset for solar flare prediction. We also handle the extreme class imbalance issue and temporal coherence issue that are exhibited in the dataset.

Committee
Dr. Rafal Angryk (chair)
Dr. Azim Ahmadzadeh