[SciPy-User] Gamma distribution in scipy.stats
Thu Jan 28 11:45:01 CST 2010
On Thu, Jan 28, 2010 at 12:36 PM, Gökhan Sever <email@example.com> wrote:
> Could someone explain to me why doesn't scipy explicitly use the location
> and scaling parameters representing its PDF?
> gamma.pdf(x,a) = x**(a-1)*exp(-x)/gamma(a) for x >= 0, a > 0.
> Gamma is represented with two or three parameters; as the function prototype
> shows correctly.
> Also both GSL and R is fine representing the Gamma PDF with two parameters:
> Should these instances need to be updated?
This should be the same if you replace `x` in the pdf by `(x-loc)/scale`
loc and scale are handled generically, while the individual _pdf
method only defines the standardized distribution.
I think I checked the algebra once for gamma, but I'm not sure.
> PS: I am wrapping the distribution definitions given in the GSL library for
> SAGE. While doing this I want to make sure my understanding and use of the
> terms and concepts are correct.
> SciPy-User mailing list
More information about the SciPy-User