categorical does not work with Pandas DataFrames

In [9]: sm.categorical(pd.DataFrame({'a':[1,2,12], 'b':['a','b','a']}), col='a')
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-9-5966a3ee6951> in <module>()
----> 1 sm.categorical(pd.DataFrame({'a':[1,2,12], 'b':['a','b','a']}), col='a')

/usr/local/lib/python2.7/dist-packages/statsmodels-0.6.0-py2.7-linux-x86_64.egg/statsmodels/tools/tools.pyc in categorical(data, col, dictnames, drop)
    143     #TODO: add a NameValidator function
    144     # catch recarrays and structured arrays
--> 145     if data.dtype.names or data.__class__ is np.recarray:
    146         if not col and np.squeeze(data).ndim > 1:
    147             raise IndexError("col is None and the input array is not 1d")

/usr/local/lib/python2.7/dist-packages/pandas-0.13.0_285_gfcfaa7d-py2.7-linux-x86_64.egg/pandas/core/generic.pyc in __getattr__(self, name)
   1802                 return self[name]
   1803             raise AttributeError("'%s' object has no attribute '%s'" %
-> 1804                                  (type(self).__name__, name))
   1805 
   1806     def __setattr__(self, name, value):

AttributeError: 'DataFrame' object has no attribute 'dtype'

an alternative is to use pd.get_dummies:

pd.get_dummies(pd.DataFrame({'a':[1,2,12], 'b':['a','b','a']}).b )

[josef-pkt]

categorical still doesn’t work with a DataFrame, which is a bit annoying for updating older code. e.g. test_glm #4432 categorical is still nice when we quickly want to get an exog without going through the formulas or the full pandas categorical type.

[bfcondon]

I have submitted a PR #3203 on this issue. It should handle DataFrames as well as other data formats.