Poster abstract details

Identifying Blazar Variability with Machine Learning
Zachary R. Weaver, Alan P. Marscher, Svetlana G. Jorstad


One of the most common characteristics of blazars is the presence of rapid variability at almost all wavelengths of the electromagnetic spectrum. Thus, variability studies are one of the most powerful tools for understanding the physical processes of blazars. There have been several statistical methods to detect variability in light curves of blazars, such as the C-test, the F-test, the $\chi^2$ test, and the analysis of variance (ANOVA) test. However, all such statistical tests require well-sampled light curves in order to determine variability. Machine learning has the capability to also detect variability for poorly-sampled light curves after being trained with well-sampled light curves. Here we present preliminary results in the use of machine learning to identify variability in the light curve of blazars, and an analysis of which light curve features are most helpful in determining variability.