Discrete uniform and hyper geometric moment

[farhanreynaldo]

Add moments and tests for the below distributions as part of #5078:

• pymc.distributions.Discrete.DiscreteUniform
• pymc.distributions.Discrete.HyperGeometric

[farhanreynaldo]

I am not quite sure why the HyperGeometric moment tests are failing, I have checked the calculation manually but it yielded a different result. Is it anything to do with the good and bad variables?

[ricardoV94]

Suggested change

	mode = intX(at.floor((n + 1) * (k + 1) / (N + 2)))
	mode = at.floor((n + 1) * (k + 1) / (N + 2))

Should be fine without intX, we take care of that in the method that calls get_moment

[farhanreynaldo]

Sure, I removed it in the latest commit.

[ricardoV94]

I am not quite sure why the HyperGeometric moment tests are failing, I have checked the calculation manually but it yielded a different result. Is it anything to do with the good and bad variables?

The get_moment should expect whatever is passed to super().dist at the end of the dist classmethod. That means get_moment is getting good, bad, and n. You will have to work with those (or convert back to N, k, n) if you need them.

[codecov]

Codecov Report

Merging #5167 (43585e1) into main (7745f55) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #5167      +/-   ##
==========================================
+ Coverage   78.05%   78.10%   +0.04%     
==========================================
  Files          88       88              
  Lines       14142    14153      +11     
==========================================
+ Hits        11039    11054      +15     
+ Misses       3103     3099       -4     

Impacted Files	Coverage Δ
pymc/distributions/discrete.py	99.42% <100.00%> (+0.01%)	⬆️
pymc/bart/pgbart.py	96.61% <0.00%> (+0.37%)	⬆️
pymc/backends/report.py	91.60% <0.00%> (+2.09%)	⬆️

[farhanreynaldo]

I am not quite sure why the HyperGeometric moment tests are failing, I have checked the calculation manually but it yielded a different result. Is it anything to do with the good and bad variables?

The get_moment should expect whatever is passed to super().dist at the end of the dist classmethod. That means get_moment is getting good, bad, and n. You will have to work with those (or convert back to N, k, n) if you need them.

So whenever we create a get_moment function, the arguments that we passed should match the arguments inside super().dist?

[ricardoV94]

I am not quite sure why the HyperGeometric moment tests are failing, I have checked the calculation manually but it yielded a different result. Is it anything to do with the good and bad variables?

The get_moment should expect whatever is passed to super().dist at the end of the dist classmethod. That means get_moment is getting good, bad, and n. You will have to work with those (or convert back to N, k, n) if you need them.

So whenever we create a get_moment function, the arguments that we passed should match the arguments inside super().dist?

Yes, that what will be used when we call get_moment internally

[farhanreynaldo]

Is the failing test related to this PR?

[ricardoV94]

Probably not, seems to be related to #4771

[ricardoV94]

Seems good to me. You have some conflicts that need to be fixed though

[ricardoV94]

Thanks @farhanreynaldo