ENH add MI simulations

[PSSF23]

Add simulations that use conditional entropy to calculate MI

[codecov]

Codecov Report

Attention: 71 lines in your changes are missing coverage. Please review.

Comparison is base (06a7066) 90.18% compared to head (b7abbd9) 88.85%.
Report is 4 commits behind head on main.

❗ Current head b7abbd9 differs from pull request most recent head bb3ec8a. Consider uploading reports for the commit bb3ec8a to get more accurate results

Files	Patch %	Lines
sktree/datasets/hyppo.py	31.06%	71 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #199      +/-   ##
==========================================
- Coverage   90.18%   88.85%   -1.33%     
==========================================
  Files          49       49              
  Lines        4198     4335     +137     
==========================================
+ Hits         3786     3852      +66     
- Misses        412      483      +71     

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

[adam2392]

Otw LGTM. I have no idea why the changelog check is failing, but we can bypass it. Same w/ style

[sampan501]

@adam2392 mix is currently an int, and realizing now should be a float, that takes values between 0 and 1. I don't think it should be a bool.

[adam2392]

Good point. Wanna add that and an error check on the value of mix being in (0,1) there?

[PSSF23]

Aren't these mu's reversed from what we defined on board? F should be mu and G should be -mu. @sampan501 @adam2392

[adam2392]

I think it should be fine since it is symmetrical.

[PSSF23]

I don't think we should add extra confusion in it.

[sampan501]

I personally don't feel strongly either way

[PSSF23]

ok

[adam2392]

Thanks @PSSF23 and @sampan501 ! Putting this in, so we can move towards some of the other tasks

If the mu0 vs mu1 ends up being confusing, feel free to propose the change in another PR.