Pandas - write dataframe in fixed-width formatted lines to a file

user_3265764 0
Viewed 747

Have a huge pandas dataframe (df) like this:

        id          date      a      b      c
0     0023  201110132120    -30    -45      7
1     0023  201110132130    -30     11   9111
2     0023  201110132140    -24     44    345
3     0023  201110132150    -19    223     11
4     0023  201110132200    -23  -3456  -1250

I need to write this dataframe to a file with special fixed-width for each field. For this i used numpy, f.e.:

np.savetxt('out.txt', df.values, fmt='%+4s %+12s %+5s %+5s %+6s')

That work's fine. Only lost header in this case. Is there a workaround?

I tested it also with pandas to_string function:

df.to_string()

But it is so slow. Why? Are there other options?

pandas dataframe python-3.x numpy python
2 Answers

One option is to abuse header option in savetxt:

formats = '%+4s %+12s %+5s %+5s %+6s'

headers = [format(str(x),y.replace('%+','>')) 
              for x, y in zip(df.columns,formats.split())]

np.savetxt('out.txt', df.values, fmt=formats,
           header=' '.join(headers), comments='')
user_42384080 
header='{:>4s} {:>12s} {:>5s} {:>5s} {:>6s}'.format('id','date','a','b','d')
np.savetxt('out.txt', df.values, fmt='%+4s %+12s %+5s %+5s %+6s', header=header)
user_145849080
Related