DataFrame.to_csv bug: headers mis-aligned

[xdong]

pandas version: 0.4.0 from pypi

The problem is that there is no header for the index column:

In [1]: import pandas

In [2]: d = pandas.DataFrame(randn(3, 2), range(3), ['a', 'b'])

In [3]: d.to_csv('tmp.csv')

In [4]: cat tmp.csv
a,b
0,-0.244543341458,-1.50542482731
1,-1.02524916375,0.279887233706
2,0.0657877607121,-0.786951623121

[wesm]

Hi,

That was actually an intentional change. If you run

read_csv('tmp.csv')

it will read it back fine. If you want the old behavior, do:

d.to_csv('tmp.csv', index_label='index')

let me know if that works and I'll close the issue

[xdong]

Oh I see. It works for me now. Thanks.

The default behavior may be somewhat misleading if the saved file is viewed as text or opened with excel, especially when there are many columns. I think that people consider csv file a table format, as such headers should be aligned with columns. Is it better to have the default output ',a,b' instead of 'a,b'?

In any case this is a minor point. It's perfectly fine the way it is. BTW great work on pandas -- it has become a indispensable tool for me.

[wesm]

Cool. Maybe, I'll take a look, so I'll leave the issue open. I was planning to be a bit more consistent with conventions in other places, like R. Will let you know

[wesm]

This behavior's been reverted in the git HEAD. I'll put out a bugfix 0.4.1 release within the next week or so or you can install the latest dev version from git.