Adapting Supervised Machine Learning to Active Galaxy Classification by Application of Multi-wavelength Data

In this study, we analyze and differentiate Active Galactic Nuclei (AGN), large bright gas regions powered by supermassive black holes, from star forming galaxies, which are galaxies that are actively producing new stars at a significant rate. Despite having physical differences, these AGN and star forming galaxies have very similar traits when it comes to their appearances,making them difficult to distinguish from one another. We formulated various models to differentiate the two, based on the ratios of elements contained within them, as revealed by astronomical spectroscopy. Our primary dataset used spectroscopy from the Sloan Digital Sky Survey, combined with photometry from ultraviolet and infrared space telescopes.The classification models we employed were K-Nearest Neighbors (KNN), Random Forest, and a Linear SVC model to determine the best possible approach for differentiating AGN from star forming galaxies. We show that our best model is as reliable as The BPT diagram, which is currently the state of the art model for differentiating AGN from star forming galaxies. This study shows that machine learning classifiers are able to be efficiently and effectively applied to multiwavelength astronomical datasets.