Add mode to Binomial

[rohanbabbar04]

Description

• Binomial

Checklist

• Code style is correct (follows ruff and black guidelines)

[rohanbabbar04]

Some things to ask here:

1. In some cases, the binomial distribution is bimodal. Should we focus on the major mode only?

2. For bimodal or multimodal like we discussed is it fine to return a tuple of modes and handle it in optimize max ent.

What do you suggest @aloctavodia ?

[Advaitgaur004]

I am adding a mode to other Python files, and I will make a pull request. is that fine @rohanbabbar04

[rohanbabbar04]

I am adding a mode to other Python files, and I will make a pull request. is that fine @rohanbabbar04

Sure, can you comment on #604 which ones are you working on so that we don't pick the same ones?

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 80.00000% with 1 line in your changes missing coverage. Please review.

Project coverage is 74.39%. Comparing base (f25da81) to head (f6a599c).
Report is 99 commits behind head on main.

Files with missing lines	Patch %	Lines
preliz/distributions/binomial.py	80.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #635      +/-   ##
==========================================
- Coverage   82.23%   74.39%   -7.85%     
==========================================
  Files         101      105       +4     
  Lines        8020     8735     +715     
==========================================
- Hits         6595     6498      -97     
- Misses       1425     2237     +812     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Advaitgaur004]

I am adding a mode to other Python files, and I will make a pull request. is that fine @rohanbabbar04

Sure, can you comment on #604 which ones are you working on so that we don't pick the same ones?

Sure, you have already implemented Rice and Binomial, so I will drop this since I have already implemented it in my local setup.

[aloctavodia]

Some things to ask here:

1. In some cases, the binomial distribution is bimodal. Should we focus on the major mode only?

2. For bimodal or multimodal like we discussed is it fine to return a tuple of modes and handle it in optimize max ent.

What do you suggest @aloctavodia ?

Not sure, but probably better to return a tuple if more than one mode. And as a first approach allow to fix a single mode in maxent, but handle de case that the mode method could return a tuple

[rohanbabbar04]

Numba vectorize doesn't support Tuple to be returned, is there a possible way to include this

def mode(self):
    y = (self.n + 1) * self.p
    if (np.isin(self.p, [0, 1])) | (np.mod(y, 1) != 0):
        return np.where(self.p == 1, self.n, np.floor(y))
    return y, y - 1

[aloctavodia]

Not sure what to do. Having to deal with multiple modes is quite annoying and I am not sure it is worth the effort.

[rohanbabbar04]

Hmm...
Should we go with the major mode instead?. We show only one mode

[aloctavodia]

How weird would be two return two modes (if they exists) only if all parameters are scalar. And return only one for vector parameters. We support vector parameters, but we mostly care of scalars

[rohanbabbar04]

I think the latest commit should work for both scalar and vectors(with scalar giving bimodal result in that case)
What do you think @aloctavodia ?