Sort pandas DataFrame with MultiIndex according to column value

Question

I have a DataFrame with MultiIndex looking like this after printing in the console:

                             value  indA  indB
           scenarioId group                        
2015-04-13    1       A           -54.0   1.0   1.0
                      B          -160.0   1.0   1.0
                      C           -15.0   0.0   1.0
              2       A           -83.0   1.0   1.0
              3       A           -80.0   2.0   2.0
              4       A          -270.0   2.0   2.0
2015-04-14    1       A           -56.0   1.0   1.0
                      B            -1.0   1.0   1.0
                      C           -60.0   0.0   1.0
              2       A           -32.0   1.0   1.0
              3       A           -91.0   2.0   2.0
              4       A           -17.0   2.0   2.0

I got it after I used the groupby and sum functions on my initial dataset.

I would like to keep the same format, but order it according to the value column. I have tried hard to do it using the sorting functions, but I think that the fact of having the first index (for the dates) of the MultiIndex without name is a problem.

Essentially, the output should look like this:

                             value  indA  indB
           scenarioId group                        
2015-04-13   1        B          -160.0   1.0   1.0
                      A           -54.0   1.0   1.0
                      C           -15.0   0.0   1.0
             2        A           -83.0   1.0   1.0
             3        A           -80.0   2.0   2.0
             4        A          -270.0   2.0   2.0
2015-04-14   1        C           -60.0   1.0   1.0
                      A           -56.0   1.0   1.0
                      B            -1.0   0.0   1.0
             2        A           -32.0   1.0   1.0
             3        A           -91.0   2.0   2.0
             4        A           -17.0   2.0   2.0

Could someone enlighten me on this please?

Thanks in advance.

jezrael · Accepted Answer · 2017-04-18 12:42:05Z

7

You can use sort_values + sort_index:

print (df.sort_values('value').sort_index(level=[0,1], sort_remaining=False))
                             value  indA  indB
           scenarioId group                   
2015-04-13 1          B     -160.0   1.0   1.0
                      A      -54.0   1.0   1.0
                      C      -15.0   0.0   1.0
           2          A      -83.0   1.0   1.0
           3          A      -80.0   2.0   2.0
           4          A     -270.0   2.0   2.0
2015-04-14 1          C      -60.0   0.0   1.0
                      A      -56.0   1.0   1.0
                      B       -1.0   1.0   1.0
           2          A      -32.0   1.0   1.0
           3          A      -91.0   2.0   2.0
           4          A      -17.0   2.0   2.0

Another solution - sort_values with reset_index and set_index:

df = df.reset_index()
       .sort_values(['level_0','scenarioId','value'])
       .set_index(['level_0','scenarioId','group'])
print (df)
                             value  indA  indB
level_0    scenarioId group                   
2015-04-13 1          B     -160.0   1.0   1.0
                      A      -54.0   1.0   1.0
                      C      -15.0   0.0   1.0
           2          A      -83.0   1.0   1.0
           3          A      -80.0   2.0   2.0
           4          A     -270.0   2.0   2.0
2015-04-14 1          C      -60.0   0.0   1.0
                      A      -56.0   1.0   1.0
                      B       -1.0   1.0   1.0
           2          A      -32.0   1.0   1.0
           3          A      -91.0   2.0   2.0
           4          A      -17.0   2.0   2.0

edited Apr 18, 2017 at 12:42

answered Apr 18, 2017 at 12:36

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

JejeBelfort Over a year ago

Maybe a quick question related to the same topic: is there a simple way to only keep the 2 smallest values from the data frame? i.e. For scenario1, to only keep group B and A on April 13th 2015 and C and A on April 14th 2015?

jezrael Over a year ago

Hmmm, is possible use df = df.groupby(level=[0,1]).head(2) but all groups (I am not sure if need it)

jezrael Over a year ago

Or maybe need df = df.groupby(level=0)['value'].nsmallest(2)

Collectives™ on Stack Overflow

Sort pandas DataFrame with MultiIndex according to column value

1 Answer 1

3 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related