How are the values sorted when I use row.values() in an graphlab.SFrame row?
For example:
import graphlab as gl
x = gl.SFrame({'X200':[1,2,3], 'X100':[4,5,6], 'X1': [7,8,9]})
row = x[0] # first row
print row.values()
[out]:
[ 7, 4, 1]
So it seems like they are sorted by the key / column names, but is that always true? Any pointers to documentation from python dictionary or SFrame will be much appreciated.
No the row.values() will not return a sorted list since row is a python dictionary and it is unsorted.
To prove the point:
>>> x = gl.SFrame({'X1': ['foo', 'bar', 'blah'], 'X3':[1,2,3], 'X2':[4,5,6]})
>>> x[0].values()[1:]
[1, 'foo']
It seems like to get a sorted values, you would have to do some key sorting first:
>>> _, v = zip(*sorted(x[0].items())[1:])
>>> v
(4, 1)
This is more of a Python dictionary issue (see Key Order in Python Dictionaries). In python, the .values() method returns values in arbitrary order. A row of the SFrame is a dictionary.
