A Random Forest spectral classification of the Gaia 500-pc white dwarf population

TL;DR

This paper utilizes a Random Forest machine learning algorithm to classify the spectral types of nearly 79,000 white dwarfs within 500 parsecs using Gaia data, significantly aiding stellar population studies.

Contribution

It introduces a novel application of Random Forest classification to Gaia spectra for white dwarfs, achieving high accuracy and creating a comprehensive spectral catalog.

Findings

Achieved 0.91 accuracy in spectral classification

High recall for DAs, DBs, DCs; good precision for DQs, DZs, DOs

Enabled better estimates of stellar evolution parameters

Abstract

The third Gaia Data Release has provided the astronomical community with astrometric data of more than 1.8 billion sources, and low resolution spectra for 220 million. Such a large amount of data is difficult to handle by means of visual inspection. In this work, we present a spectral analysis of the Gaia white dwarf population up to 500 pc from the Sun based on artificial intelligence algorithms to classify the sample into their main spectral types and subtypes. In order to classify the sample, which consists of 78 920 white dwarfs with available Gaia spectra, we have applied a Random Forest algorithm to the Gaia spectral coefficients. We used the Montreal White Dwarf Database of already labeled objects as our training sample. The classified sample is compared with other already published catalogs and with our own higher resolution Gran Telescopio Canarias (GTC) spectra, enabling the construction of a golden sample of well-classified objects. The Random Forest spectral classification of the 500-pc white dwarf population achieves an excellent global accuracy of 0.91 and an F1-score of 0.88 for the DA versus non-DA classification. In addition, we obtain a very high accuracy of 0.76 and a global F1-score of 0.62 for the non-DA subtype classification. In particular, our classification shows an excellent recall for DAs, DBs and DCs and a very good precision for DQs, DZs and DOs. The use of machine learning techniques, particularly the Random Forest algorithm, has enabled us to spectrally classify 78,920 white dwarfs with reasonable accuracy. Having an estimate of the spectral type for the vast majority of white dwarfs up to 500 pc provides the possibility of making better estimates of cooling ages, star formation rates, and stellar evolution processes, among other fundamental aspects for the study of the white dwarf population.