[SciPy-User] Any planned work on scipy.stats.distributions?

Neal Becker ndbecker2@gmail....
Wed Oct 6 08:35:42 CDT 2010


josef.pktd@gmail.com wrote:

> On Tue, Oct 5, 2010 at 7:52 AM, Neal Becker <ndbecker2@gmail.com> wrote:
>> I'm needing doubly noncentric F distribution.
>>
>> I found this:
>> http://www.ifpress.com/pdfs/comm22-897.pdf
>>
>> It might not be the best, but it's all I found.
> 
> Looks interesting, I never heard of the doubly noncentric something
> distributions.
> 
> Both your code and the paper are a bit stingy on explanations.
> 
>>From a few examples that I tried, the precision looks pretty good. I
> found some papers that use saddle point approximation to get both pdf
> and cdf of doubly non-central f and t distributions, which might be
> faster for some use cases (e.g. evaluating the cdf or pdf at many
> points.)
> 
> But for the statistics applications that I saw mentioned (needing only
> one or a few points of the cdf), your version looks good and could be
> dressed up into a scipy.stats.distributions case.
> 
> Thanks,
> 
> Josef
> 
> 

I also found
Mudholkar, Govind S., Chaubey, Yogendra P. and Ching-Chuong, Lin(1976) 
Communications in Statistics - Theory and Methods, 5: 1, 49-63
Approximations for the doubly noncentral-F distribution

I haven't tried coding it (and probably won't)



More information about the SciPy-User mailing list