Predicting Solar Flares using Support Vector Machines
The sun produces solar flares, which have the power to affect the Earth and near-Earth environment with their great bursts of electromagnetic energy and particles. These flares have the power to blow out transformers on power grids and disrupt satellite systems. As a result, we want to predict such flares to minimize its negative impact. Doing so can be a difficult because of the rarity of these events. In this iPython notebook, we explored such a challenge by extending upon the work of Bobra and Couvidat (2015).
We categorized a class of positive and negative events that correspond with flaring and non-flaring active regions on the sun. Then we created various sets of features to describe these events. Using these features, we trained and tested using a machine learning algorithm known as a Support Vector Machine and evaluated its performance using a metric known as a True Skill Score. We were able to obtain an improvement on their original work by using additional features (that quantified the maximum change in the value of certain parameters of an active region) which were shown to have strong predictive power.
|Type of resource
|Trinidad, Jacob Conrad
|Stanford Solar Observatories Group
|support vector machines
- Use and reproduction
- User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
- Preferred Citation
- Trinidad, Jacob Conrad and Bobra, Monica. (2016). Predicting Solar Flares using Support Vector Machines. Stanford Digital Repository. Available at: http://purl.stanford.edu/yv269xg0995
Stanford Research DataView other items in this collection in SearchWorks
Also listed in
Loading usage metrics...