Automatic Prediction of Solar Flares using Machine Learning: Practical Study on the Halloween Storm

In this paper, a machine learning system that can provide short-term automated prediction for the occurrences of significant solar flares is presented. This system extracts the experts´ knowledge embedded in the public NGDC solar catalogues and represents it in learning rules that can be used by computers to predict flares. This work builds on our previous work and the prediction system is tested intensively using the Jackknife technique and using real input samples from the Halloween storm. The system has managed to predict all the significant flares that took place during this storm.