Photometric Classification of Tidal Disruption Event Host Galaxies Using Machine Learning

Abstract Tidal Disruption Events (TDEs) are transient events observed when a star gets tidally destroyed by a supermassive black hole. TDEs prefer quiescent Balmer-strong (QBS) host galaxies, which have strong Balmer absorption and weak emission lines. Identifying these galaxies photometrically is difficult but important for TDE classification. This research provides a method of using photometric data to identify QBS galaxies by applying machine learning classifiers such as the Random Forest Classifier (RF) and the K-Neighbors Classifier (KN). The classifiers cumulatively identify 9548 likely QBS galaxies, with the RF achieving 75%–96% purity and 50% completeness and the KN achieving 60%–99% purity and 45% completeness for QBS galaxies below photometric redshifts of 0.3.