GALEX-SDSS photometric redshifts: Chi2
classification
We propose the following classification, based on the chi square
values obtained during the template-fitting:
|
Galaxies:
|
chipolyfit |
< |
chistar |
and chipolyfit |
< |
2chiqso |
|
Stars:
|
chistar |
< |
chipolyfit |
and chistar |
< |
2chiqso |
|
Qsos:
|
chiqso |
< |
1/2 chipolyfit |
and chiqso |
< |
1/2chistar
|
chipolyfit is the chi2 derived while fixing the
redshift at the value obtained with Polyfit. The factor 2 is empirical.
The following tables sum up the confirmation and contamination rates
of this classification by comparing the SDSS spectroscopic type and
the zphotometric one:
Confirmation rates of the chi2
classification
| Spectroscopic type | % classified as Zphotometric type |
| Galaxies | Stars | Qsos |
| Galaxies | 94.5 | 1.5 | 4 |
| Stars | 7 | 90 | 3 |
| Qsos | 24 | 1 | 75 |
|
Contamination rates of the chi2
classification
| Zphotometric type | Contamination rate (according to spectro type) |
| Galaxies | Stars | Qsos |
| Galaxies | -- | 1 | 5 |
| Stars | 10 | -- | 2 |
| Qsos | 19 | 2 | -- |
|
For instance, 94.5% of spectroscopic galaxies are classified as galaxies
by this chi2 classification, and 1.5% as stars. On the other
hand, there is 1% of spectroscopic stars and 5% of spectroscopic
qsos among the objects classified as galaxies by the chi2
classification.
In summary, the chi2 classification is efficient for
galaxies and stars, but less powerful for selecting qsos, and hence
should be taken with caution for this type. Once again, these numbers
hold for GALEX objects with SDSS spectroscopic counterparts, and are
only valid for bright objects.