[Tutor] Python OLS help

[Michael Selik]

You're welcome. A tip for the next bug: Google is pretty good at finding a
discussion of the error if you paste the whole phrase as your search terms.
In this case, I searched for "ValueError for numerical factors num_columns
must be an int" and found the relevant Google Group thread.

PS. I replied to the mailing list in case someone else has the same issue.
These emails are archived and searchable.


On Tue, May 31, 2016 at 10:07 PM Vadim Katsemba <vkatz722 at gmail.com> wrote:

> I upgraded to patsy 0.4.1 and got the regression to run. Thank you very
> much for your help!
>
> On Tue, May 31, 2016 at 8:03 PM, Michael Selik <michael.selik at gmail.com>
> wrote:
>
>> On Tue, May 31, 2016 at 5:45 PM Vadim Katsemba <vkatz722 at gmail.com>
>> wrote:
>>
>>> I typed in lm = smf.ols(formula='LATITUDE~DIAMETER',data=dataf).fit(),
>>> and I ended up getting this error: ValueError: For numerical factors,
>>> num_columns must be an int.
>>>
>>
>> You may be using an old version of Patsy, the module that allows you to
>> specify your OLS formula like R-lang does. What version of patsy are you
>> using? This seems to have been a problem with Patsy v0.4.0 and was fixed in
>> v0.4.1 (as per an email thread I read [0])
>>
>> [0] https://groups.google.com/forum/#!topic/pystatsmodels/KcSzNqDxv-Q
>>
>
>