read_csv with usecols and implicit index columns

[JamesRamm]

Pandas 0.12
When specifying usecols with read_csv it seems impossible to use an implicit (no header) index column

Example:

>>> data = 'a,b,c\n4,apple,bat,5.7\n8,orange,cow,10'
>>> pd.read_csv(StringIO(data))
        a    b     c
4   apple  bat   5.7
8  orange  cow  10.0
>>> pd.read_csv(StringIO(data), usecols=['a', 'b'])
        a    b     
0   apple  bat
1  orange  cow
>>>

Where you would prefer/expect this:

>>> pd.read_csv(StringIO(data), usecols=['a', 'b'])
        a    b     
4   apple  bat
8  orange  cow

Specifying index_col = 0 would give:

>>> pd.read_csv(StringIO(data), usecols=['a', 'b'], index_col = 0)
          b     
apple   bat
orange  cow

So this would only work if we explicity pass the column to be used as an index to usecols. However, we cannot explicitly pass an implicit index column to usecols as the index column is not named (it has no header).

[jtratner]

Are you sure you're using current master? Here's what I get for your example:

In [2]: data = 'a,b,c\n4,apple,bat,5.7\n8,orange,cow,10'

In [3]: from StringIO import StringIO

In [4]: pd.read_csv(StringIO(data))
Out[4]:
        a    b     c
4   apple  bat   5.7
8  orange  cow  10.0

In [5]: pd.read_csv(StringIO(data), usecols=['a', 'b'])
Out[5]:
        a    b
4   apple  bat
8  orange  cow
In [6]: pd.__version__
Out[6]: '0.12.0-625-g3e47a3b'

[jreback]

this was closed recently via : #5003

I believe op is using 0.12, which was buggy

[JamesRamm]

I'm using version 0.12.0 - this not the most recent?

[jreback]

it is the most recent release version. 0.13 several weeks away