Catalogue of Bayesian SZNet's spectroscopic redshift predictions

The "dr16q_superset_redshift.csv" file provides a catalogue of spectroscopic redshift predictions for spectra from the 16th data release of the Sloan Digital Sky Survey (SDSS) quasar superset catalogue (Lyke et al., 2020). Redshifts are predicted by a Bayesian convolutional neural network named Bayesian SZNet with associated predictive uncertainties in the form of predictive variances. The catalogue is released in the CSV format with the following columns:

plate: spectroscopic plate number;

mjd: modified Julian day of the spectroscopic observation;

fiberid: fiber identification number;

z_pred: redshift from Bayesian SZNet;

variance: predictive variance associated with redshift from Bayesian SZNet;

z: primary redshift;

source_z: origin of the reported redshift in z;

is_qso_final: flag indicating quasars included in the DR16Q (Lyke et al., 2020);

z_vi: redshift from visual inspection;

z_pipe: redshift from the SDSS pipeline;

zwarning: quality flag on the redshift from the SDSS pipeline;

z_dr12q: redshift from the DR12Q catalogue (Pâris et al., 2017);

z_dr7q_sch: redshift from the DR7Q catalogue (Schneider et al., 2010);

z_dr6q_hw: redshift from the DR6 catalogue (Hewett and Wild, 2010);

z_10k: redshift from the random visual inspection of 10000 spectra in the DR16Q superset;

z_pca: redshift from the redvsblue algorithm;

z_qn: redshift from QuasarNET (Busca and Balland, 2018);

z_pred_1 to z_pred_256: sampled redshifts from Bayesian SZNet;

where columns z, source_z, is_qso_final, z_vi, z_pipe, zwarning, z_dr12q, z_dr7q_sch, z_dr6q_hw, z_10k, z_pca, and z_qn are taken from the 16th data release of the SDSS quasar superset catalogue.