Re: the implementation in check_distribution_rvs

It looks like the intent is to perform a one-sample KS test with a fresh
random sample.

According to the documentation of `kstest`,
https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.kstest.html
if the first argument is the name of a distribution, a sample will be drawn
from the distribution before performing the test. When the second argument
is the name of a distribution, this distribution's CDF is used for
comparison. So to me, it looks like it's doing what it intends to do.

If the second argument of `kstest` were `rvs` here, then it would perform a
two-sample test between a new random sample from the distribution and the
original sample. That does not appear to be the intent.

Matt

On Tue, May 9, 2023, 5:01 AM <[email protected]> wrote:

> Hi,
>
> I read the code of the function check_distribution_rvs and can't
> understand the logic behind it:
>
> - in the code:
> https://github.com/scipy/scipy/blob/878285cfa3bdaf2abaa3b8f7e51ae4a277d63b33/scipy/stats/tests/test_continuous_basic.py#LL664C5-L664C27
> - in L678: why is "stats.kstest(dist, dist, args=args, N=1000)"? I thought
> it should be "stats.kstest(dist, rvs, args=args, N=1000)"?
>
> because the comments also say "Repeat the test with a new sample of rvs.
> Generate 1000 rvs, perform a K-S test that the new sample of rvs". Did I
> miss something? Thanks.
> _______________________________________________
> SciPy-Dev mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> https://mail.python.org/mailman3/lists/scipy-dev.python.org/
> Member address: [email protected]
>
>

_______________________________________________
SciPy-Dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3/lists/scipy-dev.python.org/
Member address: [email protected]