[SciPy-dev] GLMs ?

josef.pktd@gmai... josef.pktd@gmai...
Fri Aug 14 18:29:04 CDT 2009


On Fri, Aug 14, 2009 at 5:46 PM, Pierre GM<pgmdevlist@gmail.com> wrote:
>
> On Aug 14, 2009, at 5:27 PM, Skipper Seabold wrote:
>
>> On Fri, Aug 14, 2009 at 5:19 PM, Pierre GM<pgmdevlist@gmail.com>
>> wrote:
>>> All,
>>> What's the status of GLMs in scipy ? Does anybody have some code they
>>> could point me to ?
>>> Thanks a lot in advance
>>> P.
>>
>> GLMs got a lot of attention as part of my GSoC.  You can find the
>> (still in flux) code here
>>
>> <http://bazaar.launchpad.net/~scipystats/%2Bjunk/skipper-working/files/head%3A/nipy/fixes/scipy/stats/models/
>> >
>
> Fab'.
> FYI, I need to fit Tweedie distributions to precipitation series. I
> have already coded the distributions in the scipy standard, and now I
> need to estimate the parameters...
> Thanks again
> P.
> _______________________________________________
> Scipy-dev mailing list
> Scipy-dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-dev
>

What's GLM ? Generalized, not General Linear Model, just to make sure
we are talking about the same thing. I don't see how GLM can help you
directly to estimate the parameters of a distribution, unless it is
one of the exponential family distributions covered with a standard
Generalized Linear Model implementation.

I don't know the Tweedie distribution, but from the reference Skipper
gave you, you can see GLM covers gaussian, binomial, gamma and a few
others. The main point of GLM is to regress the mean/expectation of
the distribution on some explanatory variables.

Are you trying to estimate parameters of the distribution themselves,
or parameters of the distribution as function of some explanatory
variables? In the first case, GLM won't be of much help.

Josef


More information about the Scipy-dev mailing list