Signal to Noise Cuts on Sample Selection
-
by
mzevin1
scientist, moderator
Hey all,
It's been a while since I've been able to post - I hope all are doing well!
In the past there was some question as to what Signal to Noise cuts should have been used for the sample. When Yanmei derived the sample, she used S/N gt 10 as a cut, and there was some speculation that S/N gt 15 would possibly result in a cleaner sample.
To test this out, I started by recreating Yanmei's sample. The relevant sample selection files can be found at http://quenchtalk.galaxyzoo.org/#/boards/BGS0000001/discussions/DGS00001xy. Using the principal component analysis method, this sample was cut to about 14,000 sources (the principal component criteria are as follows: sel = where{pc2 gt a*pc1+b and pc1 lt -1.8} with a=-0.146667 and b=0.33332).
From here, duplicate sources were removed and I made two separate cuts to the sample, once using S/N gt 10 and again using S/N gt 15. When using the restriction S/N gt 10, I was left with about 3100 sources, and with S/N gt 15 about 1200 sources. I proceeded to graph stellar masses vs redshift of these sources:
The distributions of the two were nearly identical (other than the S/N gt 10 graph being a denser distribution), leading me to initially believe that applying S/N gt 15 would not cause too much of a change in our project's results. Let me know your thoughts on the S/N cuts!
Next I'm going look at merger fraction distributions for the to S/N cuts and cut the sample down again by restricting the redshift to z=0.02-0.08 and the absolute z-band magnitude to Abs_Mag_Z lt -20 to see what that give us. More to come!
Posted
-
by
mlpeck
I suspect not all of your post is being displayed. The Talk software thinks the symbols for "less than" and "greater than" are html tags, and stuff tends to disappear between "less than" and the next "greater than". If you want them to display properly use the html for the symbols.
As for the substance, the redshift and magnitude cuts we've already agreed to should probably eliminate the lowest S/N spectra from the sample, although that can certainly be checked.
Personally I am OK with trying to do some science with the sample we have. And if much more time is spent on sample selection I am going to have to suspend my participation until the rest of you get the sample sorted out.
Posted
-
by
JeanTate
in response to mlpeck's comment.
Hi mzevin1, long time no see! Good to see you're back posting here.
[mlpeck]: I suspect not all of your post is being displayed.
I agree. I think it's very likely that significant parts of your post have vanished from sight, due to your use of the > and/or <
symbols, in native form (the HTML symbols I use are & g t ; and & l t ; - without the spaces, of course).The good news is that the text string which comprises your post is still there, and you can edit it, to replace the native symbols with the appropriate HTML codes. 😃
Posted