This page contains links to files generated by Nick Ball for various astronomical research projects using machine learning algorithms. The archive is hosted as a community project of the Canadian Astronomical Data Centre at the Canadian National Research Council Herzberg Institute of Astrophysics. The datasets, links, explanations, and usefulness will hopefully grow in time (or get added to a proper database).
SDSS P(G,N,S) Object Classifications and Photometric RedshiftsProbabilistic classifications of 143 million Sloan Digital Sky Survey Data Release 3 objects as galaxy, star, or neither as described in Ball et al. (2006, ApJ 650 497), with the caveats therein.
The original catalogue has been split into several smaller files.
The files are still large!
They are not gzipped because on a modern network this will generally take much longer than transferring the file.
It is not necessary to download all files to use the classifications; they are provided for convenience.
The columns are formatted as ASCII csv to 5 decimal places (%.5f), or as retrieved from the SDSS database.
For more information, e.g., the meaning of 'model magnitude', see the SDSS database schema.
dr3_Oct_catalog_id.csv (2.6G) - SDSS DR3 catalogue ID (skyVersion,run,rerun,camCol,field,obj)
dr3_Oct_catalog_radec.csv (2.5G) - J2000 right ascension and declination
dr3_Oct_catalog_pgns.csv (3.2G) - Classification probability P(galaxy,neither,star)
dr3_Oct_catalog_photoz.csv (1.1G) - Photometric redshift
dr3_Oct_catalog_dered_iz.csv (2.4G) - Extinction-corrected model i and z magnitudes
dr3_Oct_catalog_dered_ugr.csv (3.6G) - Extinction-corrected model u, g, and r magnitudes
dr3_Oct_catalog_err_iz.csv (2.1G) - Model i and z magnitude errors
dr3_Oct_catalog_err_ugr.csv (3.2G) - Model u, g, and r
dr3_Oct_catalog_extinction.csv (1.1G) - Extinction due to Galactic dust, from the maps of Schlegel, Finkbeiner and Davis (1998)
Etc.Add some more interesting datasets here.
Last modified: Feb 24th 2010