How to convert each row of a Pandas DataFrame into a new nxm matrix?

Question

Hi I have generated a dataframe as follows:

Data     mS1  nS1   mS1S2   nS1S2
KC        1    9     1       18
KN        1    9     0       19
KD        1    9     1       18
NG        0    10    2       17

Now I want to convert each row as to a new 2x2 matrix: 1st & 3rd and 2nd & 4th columns

For example for KC:

 KC  1  1
-KC  9  18

Answer 1

Create a dict or list comprehension, with with keys from your Data field, and use numpy.ndarray.reshape. Looks like it's also necessary to use order='F' for desired output:

d = {i: pd.DataFrame(r.to_numpy().reshape(2, 2, order='F'),
                     index=[f'{i}', f'-{i}'],
                     columns=['A', 'B'])
     for i, r in df.set_index('Data').iterrows()}

print(d['KC'])

[out]

     A   B
KC   1   1
-KC  9  18

---

l = [pd.DataFrame(r.to_numpy().reshape(2, 2, order='F'),
                  index=[f'{i}', f'-{i}'],
                  columns=['A', 'B'])
     for i, r in df.set_index('Data').iterrows()]

for d in l:
    print(d)

[out]

     A   B
KC   1   1
-KC  9  18

     A   B
KN   1   0
-KN  9  19

     A   B
KD   1   1
-KD  9  18

      A   B
NG    0   2
-NG  10  17

Answer 2

Criteria are not clear, but a solution for this toy problem could be:

import pandas as pd

df = pd.DataFrame({'mS1':[1,1,1,0],
                   'nS1':[9,9,9,10],
                   'mS1S2':[1,0,1,2],
                   'nS1S2':[18,10,19,17]},
                  index = ['KC', 'KN', 'KD', 'NG'])

d = {}
for r in df.itertuples():
    d[r[0]] = pd.DataFrame({'A':[r[1],r[3]],
                            'B':[r[2],r[4]]},
                           index = ['{}'.format(r[0]), '-{}'.format(r[0])])

d

and you get a dictionary d with 4 dataframes:

{'KC':      A   B
       KC   1   9
      -KC   1  18,
 'KN':      A   B
       KN   1   9
      -KN   0  10,
 'KD':      A   B
       KD   1   9
      -KD   1  19,
 'NG':      A   B
       NG   0  10
      -NG   2  17}