Python Pandas provides two methods for sorting DataFrame :
- sort_values (or DEPRECATED sort)
- sort_index
What are differences between these two methods ?
Asked
Active
Viewed 2.0k times
27
nbro
- 15,395
- 32
- 113
- 196
working4coins
- 1,997
- 3
- 22
- 30
-
9They are the same. ``Sort`` is an older version that accepts a slightly different calling convention. Use ``sort_index``. – Jeff Oct 12 '13 at 12:11
-
Is there a way to sort first using index, and after sort using a column name ? – working4coins Oct 12 '13 at 12:28
-
make the index a column, then ``sort_index(by=['index_name','column_name')`` – Jeff Oct 12 '13 at 12:54
2 Answers
22
As the question was updated to ask for the difference between sort_values (as sort is deprecated) and sort_index, the answer of @mathdan is no longer reflecting the current state with the latest pandas version (>= 0.17.0).
sort_valuesis meant to sort by the values of columnssort_indexis meant to sort by the index labels (or a specific level of the index, or the column labels whenaxis=1)
Previously, sort (deprecated starting from pandas 0.17.0) and sort_index where indeed almost identical (both methods could sort by both columns and index). But this confusing situation has been solved in 0.17.0.
For an overview of the changes in the sorting API, see http://pandas.pydata.org/pandas-docs/stable/whatsnew/v0.17.0.html#changes-to-sorting-api
joris
- 133,120
- 36
- 247
- 202
-
It looks like the link that works now is: http://pandas.pydata.org/pandas-docs/stable/whatsnew/v0.17.0.html#changes-to-sorting-api – Mike Mar 03 '19 at 01:06
6
The difference is entirely in the way it is called. The source code for sort is literally a one-line call to sort_index.
