Add nu parametrization to beta distribution #6344
Conversation
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## main #6344 +/- ##
==========================================
- Coverage 94.72% 86.04% -8.69%
==========================================
Files 148 148
Lines 27487 27818 +331
==========================================
- Hits 26038 23936 -2102
- Misses 1449 3882 +2433
|
pymc/distributions/continuous.py
Outdated
| @@ -1059,10 +1059,11 @@ class Beta(UnitContinuous): | |||
| Support :math:`x \in (0, 1)` | |||
| Mean :math:`\dfrac{\alpha}{\alpha + \beta}` | |||
| Variance :math:`\dfrac{\alpha \beta}{(\alpha+\beta)^2(\alpha+\beta+1)}` | |||
| Sample size :math:`\alpha + \beta` | |||
There was a problem hiding this comment.
I am not sure about this one. It's not really a property of the distribution, but an interpretation of the parameters, assuming a specific model
There was a problem hiding this comment.
You can add this as a comment in the docstring, but I agree that this is not a property per se.
There was a problem hiding this comment.
I was hesitating as well, I will remove the reference to "sample size" without the full explanation. Should I add:
'nu : tensor_like of float, optional
Alternative "sample size" of a Beta distribution for the Haldane prior probability Beta(0,0) (nu > 0) .
in the parameter section?
There was a problem hiding this comment.
I think the parameter description is good enough as it is now.
We can perhaps add a link to the relevant Wikipedia section instead of mentioning the Haldane prior?
https://en.m.wikipedia.org/wiki/Beta_distribution#Mean_and_sample_size
|
Pre-commit is failing, otherwise the PR looks good to me |
* Add (mu, nu) parametrization * add test beta logp * Add relationship between alpha, beta and nu * change test structure * remove white space accidently added in docstring * remove sample size from docstring --------- Co-authored-by: Michael Osthege <michael.osthege@outlook.com>
What is this PR about?
...
This PR addresses closes #5023 and closes #6295 and incorporates the mu, nu parametrization to the Beta distribution as described here https://en.wikipedia.org/wiki/Beta_distribution as well as fixing the error on the sigma bounds in the docstring.
As for the test, I implemented one, on the logp, and it passed. I assumed that if logp was the same across parametrization, cdf and others would follow. Let me know if this is sufficient.
Checklist
Major / Breaking Changes
Bugfixes / New features
Betadistribution can now be parametrized bynu.Docs / Maintenance