Zooite vote distribution, two Gaussians; why?
-
by
JeanTate
Take ~30k objects in Table 2 of the Galaxy Zoo 1 data release.
Extract the number of votes for each (from the column "NVOTE"). Plot the frequency of each NVOTE; there are eight objects for which NVOTE=14; 386 for which NVOTE=40; etc.
That's the blue diamonds. Looks like two Gaussians. Simply by eyeballing this - no analysis, no apps, etc - and a little bit of trial and error, two Gaussians fits (the orange line) ... very well! 😮 8)
Why?
What is it about the original Galaxy Zoo that would lead you to a) expect there to be two Gaussians, and b) why is the apparent fit so good?
By way of comparison, that's ~10k objects, from Table 6. Only one Gaussian (again, this is a 'fit by eye', not a formal analysis).
There are two oddities here: there are seven objects with NVOTE well beyond the expected (right-side) tail, and the peak is messy.
The seven don't seem special in any way: (see the Zooite vote distribution, two Gaussians; why? thread, in GZ forum for images)
Posted
-
by
v3nom
Hi JeanTate ,
Is it number of votes on Y axis and number of days on X axis?
Quite interesting.Posted
-
by
JeanTate
Hi v3nom,
The x-axis is NVOTE, the number of zooite votes for a particular object (spiral galaxy, in this case).
The y-axis is the number of objects (spiral galaxies) with NVOTE zooite classifications.
So, for example, in the first graph/chart/plot, for NVOTE=14 (the x-axis), there are eight objects (the y-axis); for NVOTE=40, 386 objects.
Posted