Groupin by on multiple columns in python

Asked May 09 '18 at 15:23

Active May 09 '18 at 15:41

Viewed 33 times

I have a panda dataframe which has the following columns, year, month, source and some other field. I would like to group by year, month and source I want the `source`` to appear as a column, after groupby.

           source1, source2, source3,....
2017, jan
2017, feb
...

This is how my code looks like

df.pivot_table(index = ['year','month'], columns = 'source', aggfunc = np.size, fill_value = 0)

but I'm getting the following error message

Reindexing only valid wit uniquely valued index objects

edited May 09 '18 at 15:41

asked May 09 '18 at 15:23

HHH

6,085
20
92
164

show us some data ? – BENY May 09 '18 at 15:27
Can you provide a dataframe with example data? I understand the problem but, honestly, am too lazy to mock up a dataframe from scratch. – jpp May 09 '18 at 15:27
https://stackoverflow.com/questions/47152691/how-to-pivot-a-dataframe – BENY May 09 '18 at 15:29
you can always use `df.melt()` and `df.pivot_table()` – It_is_Chris May 09 '18 at 15:39
I revised my original question, could you have a look – HHH May 09 '18 at 15:41
@H.Z. here is a similar question I answered. It may guide you in the right direction: https://stackoverflow.com/questions/49539325/how-to-pivot-table-so-the-columns-can-also-be-part-of-the-index/49559384#49559384 – It_is_Chris May 09 '18 at 15:50
what should I put as the 'values' in the pivot_table? – HHH May 09 '18 at 15:58
@H.Z. without seeing the data I am guessing you dont need it. you can remove `value_name` from `df.melt` and remove `values` from `df.pivot_table` – It_is_Chris May 09 '18 at 16:07
do I have to use melt as well? Can't I just use pivot_tale? – HHH May 09 '18 at 16:11
should I first do a group by and then use the pivot_table? – HHH May 09 '18 at 17:03

Groupin by on multiple columns in python

0 Answers0