Pivot columns while retaining original column headers

Question

I would like to count number of yes and no values by Column and groupby index.

I have this dataframe :

col0  col1 col2
A     yes  no
A     no   no
B     yes  yes
B     yes  no

I want this:

   col1     col2
   yes  no  yes  no
A  1    1   0    2
B  2    0   1    1

I tried with df.pivot_table(index='my_index', aggfunc='count') but i only got

   col1     col2

A  2        2
B  2        2

@Wen, I've tried a few pivot solutions, but they don't seem to work. Any ideas? — cs95
– cs95, Commented Mar 30, 2018 at 17:58
@cᴏʟᴅsᴘᴇᴇᴅ yep pd.concat([pd.crosstab(df.col0,[df.col1.astype('category')]),pd.crosstab(df.col0,[df.col2.astype('category')])],axis=1,keys=['col1','col2']) — BENY
– BENY, Commented Mar 30, 2018 at 18:01
@Wen Ah, it's definitely more complicated than a simple pivot problem... I'll reopen this question, so please post that as the answer ;) — cs95
– cs95, Commented Mar 30, 2018 at 18:02
@cᴏʟᴅsᴘᴇᴇᴅ Yep , I think so , we should reopen it — BENY
– BENY, Commented Mar 30, 2018 at 18:03

cs95 · Accepted Answer · 2018-03-30 18:13:48Z

3

Option 1
pd.get_dummies + groupby + sum

v = pd.get_dummies(df.set_index('col0'))

v.columns = pd.MultiIndex.from_tuples(
    list(map(tuple, v.columns.str.split('_')))
)
v.sum(level=0)

     col1     col2    
       no yes   no yes
col0                  
A       1   1    2   0
B       0   2    1   1

Option 2
stack + get_dummies + unstack

(df.set_index('col0')
   .stack()
   .str.get_dummies()
   .sum(level=[0,1])
   .unstack(-1)
   .swaplevel(0, 1, axis=1)
   .sort_index(level=0, axis=1)
)

     col1     col2    
       no yes   no yes
col0                  
A       1   1    2   0
B       0   2    1   1

Option 3
crosstab + concat by @Wen

i = pd.crosstab(df.col0, df.col1.astype('category'))
j = pd.crosstab(df.col0, df.col2.astype('category'))

pd.concat([i, j], axis=1, keys=['col1','col2'])

     col1     col2    
col1   no yes   no yes
col0                  
A       1   1    2   0
B       0   2    1   1

edited Mar 30, 2018 at 18:13

answered Mar 30, 2018 at 18:04

cs95

406k106 gold badges744 silver badges795 bronze badges

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

8 Comments

BENY Over a year ago

Man just add it to your answer, without your reminder I almost killing a good question ..

pd.concat([pd.crosstab(df.col0,[df.col1.astype('category')]),pd.crosstab(df.col0,[df.col2.astype('category')])],axis=1,keys=['col1','col2'])

cs95 Over a year ago

@Wen I didn't see your answer, please post again so I can upvote ;)

BENY Over a year ago

If you do not mind man, could you please adding to your answer ? feel embarrassed to post as an answer...:-(

cs95 Over a year ago

@Wen It is a good answer, shame you did not post it yourself. I've added it, cheers

ambigus9 Over a year ago

@cᴏʟᴅsᴘᴇᴇᴅ thanks! It works perfectly! Should I change the title of Question?, Wich is a good tittle for this problem?

|

Collectives™ on Stack Overflow

Pivot columns while retaining original column headers

1 Answer 1

8 Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

8 Comments

Linked

Related