df.sortlevel() vs df.sort_index()

[amelio-vazquez-reina]

Q1: For a multi-index dataframe, what is the difference between sortlevel() and sort_index()?

As far as I can tell, for a dataframe like:

  df = pd.DataFrame({'A' : [np.random.random_integers(10) for x in xrange(500)], 
                     'B' : [np.random.random_integers(10) for x in xrange(500)], 
                     'C' : [np.random.random_integers(10) for x in xrange(500)],
                     'data' : randn(500) })

the result is the same.

Q2: Why do these methods not use the same naming convention? sort_index uses underscore to separate words, while sortlevel is just one word.

What is the official convention in Pandas? Isn't the Python convention to use underscore splitting?

[jreback]

related #5190

[jreback]

these are equivalent in this case.

sort_index is more general as it can sort by the index OR by a set of columns

sort_index could subsume sortlevel functionaility if we added a level argument.

I think that sortlevel was originally only for a multi-index and never got integrated into sort_index. as for the naming, don't know.

I'll add these comment to the issue I link which is a discussion on revising / unifying the sort API in 0.15