Why DataFrame.sort_index retuns None?

Question

I have tried to sort data frame by index and have None as result. Why it happens? This is my data frame http://screencast.com/t/zOEuR6Uu And I call:

levelData.sort_index(inplace = True)

None (((

 levelData.info()
<class 'pandas.core.frame.DataFrame'>
Index: 525 entries, 1 to 99
Data columns (total 2 columns):
win       525 non-null int64
action    525 non-null int64
dtypes: int64(2)
memory usage: 12.3+ KB

I'm guessing that it gets sorted in place ... Since it's mutating the original object (`levelData`), there is nothing to return that you don't already have an easy handle on. — mgilson, Jun 06 '16 at 16:53
@OleksandraK, and levelData.head() And with head edit new table to show the sort you want. — Merlin, Jun 07 '16 at 22:32

score 4 · Answer 1 · edited May 23 '17 at 10:29

DataFrame.sort_index is following the Command-Query separation principle which says that a function should either be a command or a query but not both.

Commands do something such as modify the caller (e.g. the DataFrame) and should return None. (In languages that allow it, commands should return nothing. But since all functions return something in Python, the convention is to return None.)
Queries compute and return a result but do not modify the caller.

inplace=True makes sort_index a command. The method modifies the calling DataFrame and in keeping with the Command-Query principle, returns None.

inplace=False (the default) makes sort_index a query. Thus, if you want sort_index to return a DataFrame, remove the inplace=True.

Truly "inplace" operations like list.sort modify the original data without using auxilliary data structures. In this sense, no Pandas operation is truly inplace. When inplace=True is used, the result is first computed in an auxilliary data structure (e.g. another DataFrame), and then the underlying data is copied into the shell of the original DataFrame. So

df.sort_level(inplace=True)

is equivalent to

df = df.sort_level()

inplace=True exists mainly because of the inertia of history. The lead developer of Pandas, Jeff Reback says:

My personal opinion: I never use in-place operations. The syntax is harder to read and its does not offer any advantages.

when I remove inplace=True data frame does not change after sort_index — OleksandraK, Jun 07 '16 at 09:10
@OleksandraK: Of course it doesn't. The new dataframe is returned from the function instead in this case *as stated very clearly at the start of this answer*. — SiHa, Jun 07 '16 at 11:09

Why DataFrame.sort_index retuns None?

1 Answers1