Should Series.__contains__ look at indexes or values?

[danielballan]

When writing this SO answer, I came upon this behavior, which I did not expect.

In [1]: 1 in Series([1])
Out[1]: False

In [2]: 1 in Series([1, 2])
Out[2]: True

Apparently, __contains__ looks at the index.

In [3]: 1 in Series([1], index=[1])
Out[3]: True

I can see how that makes sense. But I think it's fair to say that it is potentially confusing. Do we rely on this behavior of __contains__ for data alignment? Might it be reasonable to change it? I'm not sure that's a good idea, but I want to raise the issue.

[jreback]

IIRC this is for dict compat, IOW a series is dict like with the index being the keys and the values the actual values. So its consistent. For a frame, the __contains__ works on the columns axis.

isin is the way to example 'in' for values

[TomAugspurger]

This may be a useful addition to http://pandas.pydata.org/pandas-docs/dev/gotchas.html

I don't know if anyone actually reads that, but at least it would be documented.

[jreback]

what, people don't read the docs???

:)

[jreback]

@danielballan want to do an addition to the gotchas (and/or section on Series is dict-like in basics.rst)