Improved source classification and performance analysis using Gaia DR3

TL;DR

This paper enhances Gaia DR3 source classification by developing new probabilistic methods and combining classifiers, significantly improving purity especially for faint quasars, while analyzing performance variations across sky regions.

Contribution

It introduces a new additive classifier combination and a variable prior, leading to improved purity and completeness in Gaia DR3 extragalactic source classification.

Findings

Achieved 55% and 89% purity for quasars and galaxies with Combmod.

Significant purity increase for faint quasars from 20% to 62%.

Enhanced classification performance by combining classifiers and using a variable prior.

Abstract

The Discrete Source Classifier (DSC) provides probabilistic classification of sources in Gaia Data Release 3 using a Bayesian framework and a global prior. The DSC Combmod classifier in GDR3 achieved for the extragalactic classes (quasars and galaxies) a high completeness of 92%, but a low purity of 22% due to contamination from the far larger star class. However, these single metrics mask significant variation in performance with magnitude and sky position. Furthermore, a better combination of the individual classifiers is possible. Here we compute two-dimensional representations of the completeness and the purity as function of Galactic latitude and source brightness, and also exclude the Magellanic Clouds where stellar contamination significantly reduces the purity. Reevaluated on a cleaner validation set and without introducing changes to the published GDR3 DSC probabilities themselves, we achieve for Combmod average 2D completenesses of 92% and 95% and average 2D purities of 55% and 89% for the quasar and galaxy classes, respectively. Since the relative proportions of extragalactic objects to stars in Gaia is expected to vary significantly with brightness and latitude, we introduce a new prior as a continuous function of brightness and latitude, and compute new class probabilities. This variable prior only improves the performance by a few percentage points, mostly at the faint end. Significant improvement, however, is obtained by a new additive combination of Specmod and Allosmod. This classifier, Combmod-$\alpha$, achieves average 2D completenesses of 82% and 93% and average 2D purities of 79% and 93% for the quasar and galaxy classes, respectively, when using the global prior. Thus, we achieve a significant improvement in purity for a small loss of completeness. The improvement is most significant for faint quasars where the purity rises from 20% to 62%.