Pandas: Reshaping dataframe

Question

I have a panda's related question. My dataframe looks something like this:

  id val1 val2
0  1     0    1
1  1     1    0
2  1     0    0
3  2     1    1
4  2     1    1
5  2     1    0
6  3     0    0
7  3     0    1
8  3     1    1
9  4     1    0
10 4     0    1
11 4     0    0

I want to transform it into something like:

             a         b        c
   id     a0   a1   b0   b1   c0   c1
    1     0    1    1    0    0    0
    2     1    1    1    1    1    0
    3     0    0    1    1    1    1
    4     1    0    0    1    0    0

I thought of something like adding a sub_id column that is enumerated cyclically by a, b and c and then do an unstack of the frame. Is there an easier/smarter solution?

Thanks a lot!

Tim

Answer 1

If possible numbers instead abc is use GroupBy.cumcount for counter, create MultiIndex by DataFrame.set_index and reshape by DataFrame.unstack and last sorting second level with DataFrame.swaplevel :

g = df.groupby('id').cumcount()

df = df.set_index(['id', g]).unstack().sort_index(axis=1, level=1).swaplevel(0,1,axis=1)
print (df)
      0         1         2     
   val1 val2 val1 val2 val1 val2
id                              
1     0    1    1    0    0    0
2     1    1    1    1    1    0
3     0    0    0    1    1    1
4     1    0    0    1    0    0

If want a,b,c values is possible generate dictionary from string.ascii_lowercase and rename columns:

import string

d = dict(enumerate(string.ascii_lowercase))
df = df.rename(columns=d)
print (df)
      a         b         c     
   val1 val2 val1 val2 val1 val2
id                              
1     0    1    1    0    0    0
2     1    1    1    1    1    0
3     0    0    0    1    1    1
4     1    0    0    1    0    0

Solution for rename both levels is first create default columns names by range after set_index :

g = df.groupby('id').cumcount()
df = df.set_index(['id', g])
df.columns = range(len(df.columns))
df = df.unstack().sort_index(axis=1, level=1).swaplevel(0,1,axis=1)
print (df)
    0     1     2   
    0  1  0  1  0  1
id                  
1   0  1  1  0  0  0
2   1  1  1  1  1  0
3   0  0  0  1  1  1
4   1  0  0  1  0  0

And last in list comprehension set new values:

import string

d = dict(enumerate(string.ascii_lowercase))
df.columns = pd.MultiIndex.from_tuples([(d[a], f'{d[a]}{b}') for a, b in df.columns])
print (df)
    a     b     c   
   a0 a1 b0 b1 c0 c1
id                  
1   0  1  1  0  0  0
2   1  1  1  1  1  0
3   0  0  0  1  1  1
4   1  0  0  1  0  0

Answer 2

One of possible solutions:

Start from reformatting values for each id into a single row:

res = df.set_index('id').groupby('id').apply(
    lambda grp: pd.Series(grp.values.flatten()))

For now the result is:

    0  1  2  3  4  5
id                  
1   0  1  1  0  0  0
2   1  1  1  1  1  0
3   0  0  0  1  1  1
4   1  0  0  1  0  0

Then set proper column names:

res.columns = pd.MultiIndex.from_tuples(
    [(x, x + y) for x in list('abc') for y in list('01')])

The finale result is:

    a     b     c   
   a0 a1 b0 b1 c0 c1
id                  
1   0  1  1  0  0  0
2   1  1  1  1  1  0
3   0  0  0  1  1  1
4   1  0  0  1  0  0