使用熊猫拆分几列

Question

I want to split string in several columns. For example, I'd like to select some information from col2, col3 and col5 in below dataframe (but indeed I have more than hundred columns to do so).

d = pd.DataFrame({
                  'col1' : ['USA', 'AGN'],
                  'col2' : ['0|0:0.014:0.986,0.013,0', '1|0:0.02:1.936,0.023,1'],
                  'col3' : ['1|0:0.024:0.9,0.01345,2', '0|2:0.213:0.92,0.1,2'],
                  'col4' : ['done', 'done'],
                  'col5' : ['2|0:0.02:1.936,0.023,1', '1|0:0.024:0.9,0.01345,2']
                  })

  col1                     col2                     col3  col4 .....
0  USA  0|0:0.014:0.986,0.013,0  1|0:0.024:0.9,0.01345,2  done .....  
1  AGN   1|0:0.02:1.936,0.023,1     0|2:0.213:0.92,0.1,2  done .....  

I only need first 3 marks from that long string. Then I expect I can see from my result such as below.

col1 col2  col3  col4  col5  ....
USA   0|0   1|0  done   2|0  ....
AGN   1|0   0|2  done   1|0  ....

Any hint please?

Answer 1

if i understood your question correctly, you can do it this way:

In [254]: d.replace(r':.*', '', regex=True)
Out[254]:
  col1 col2 col3  col4 col5
0  USA  0|0  1|0  done  2|0
1  AGN  1|0  0|2  done  1|0

Answer 2

To get the first three string characters:

>>> d.col2.str[:3]
0    0|0
1    1|0
Name: col2, dtype: object

To split on ":" and take the first item:

>>> d.col2.str.split(':', expand=True)[0]
0    0|0
1    1|0
Name: 0, dtype: object

To apply it to a group of columns:

cols = ['col2', 'col3', 'col5']
d.loc[:, cols] = d.loc[:, cols].apply(lambda s: s.str[:3])

>>> d
  col1 col2 col3  col4 col5
0  USA  0|0  1|0  done  2|0
1  AGN  1|0  0|2  done  1|0