software-store-developers team mailing list archive

[Michael Vogt <mvo@xxxxxxxxxx>]

On Thu, Jun 23, 2011 at 10:51:22AM +1000, Aaron Peachey wrote:
> On Thu, Jun 23, 2011 at 3:34 AM, Matthew Paul Thomas <mpt@xxxxxxxxxxxxx> wrote:
> > (I would like to be able to include, in the specification, test cases
> > with expected numeric results that weren't calculated by the actual USC
> > implementation. But I have neither a book of statistical tables to
> > calculate it by hand, nor the ability to figure out quickly how to
> > install the statistics2 Ruby gem to try the sample code on the above Web
> > page. So if anyone would like to help out with that too, that would be
> > great.)
> >
> 
> I had a look around to see if I could find a python implementation of
> pnormaldist but didn't run into any luck.
> 
> So I've used the ruby source to re-write the ruby
> statistics2.pnormaldist function in python so we can easily utilise
> this instead of having to rely on the ruby version.
> 
> The file pnormaldist.tar.gz available here (http://db.tt/Bx9hmu0) has
> two scripts in it.
> pnormaldist.py - my python re-write and a main function to print the
> output of pnormaldist with values of 0.1 to 0.9 in 0.1 steps
> pnormaldist.rb. - the original statistics2.pnormaldist ruby function
> and print statements to print the same output
[..]

Thanks! Great that you took the time to convert it. I had a quick look
at this ruby statistics2 package license, but I could find
anything. It would be good to clarify the license. I don't actually
think that our version falls strictly under "derived work" as its a
implementation of a mathematical algorithm (and its pretty short) so
there is something like a obvious way of doing it (which is given by
the math). But if the license is something like MIT or even public
domain that would make this a total no-problem.
 
> Run like this:
> python pnormaldist.py
> ruby pnormaldist.rb
> 
> to confirm the function works as expected. Change the values in the
> print functions as required to test other arguments.
[..]

Ideally the ruby package would have a testsuite with some numerical
values to test against, but apparently that is not the case, the only
thing seems to be
http://blade.nagaokaut.ac.jp/~sinara/ruby/math/statistics2/statistics2-0.53/README
:

Example:
      require "statistics2"
      puts Statistics2.normaldist(0.27) #=> 0.606419873198039


Cheers,
 Michael