-
Notifications
You must be signed in to change notification settings - Fork 222
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add negative binomial distribution #356
base: master
Are you sure you want to change the base?
Add negative binomial distribution #356
Conversation
@alejandroschuler Any tips on how to fix the github actions failing? |
@ryan-wolbeck can probably help! |
@ashler-herrick you will need to install Make in your development environment then run 'make lint' that will fix the formatting error you are seeing. |
@ryan-wolbeck Any ideas on how to fix the failing tests? I've fixed the formatting errors, ran 'make test' locally (passed) but it seems to be failing for Python 3.9 through GH Actions |
@ashler-herrick it's probably failing only in the github action tests because your local environment is probably not 3.9. I would try creating a new conda env and setting it to 3.9 then do 'make' to install and re-run the tests. I'll pull the code here and take a look and get back to you with what I see. |
Interesting, when I run on 3.9 locally don't get the failure, this will take me more time to dig into. Going to also re-run the github actions to see if it's repeating |
@ashler-herrick now when I re-run the actions I get the following linting error ************* Module ngboost.api Your code has been rated at 9.86/10 ************* Module ngboost.ngboost |
I've spent some time looking into this PR again, could you move the Negative Binomial tests into test_distns.py instead of test_score.py? Since Negative Binomial is a distribution, we keep all distribution tests together for consistency (e.g., like Normal, Laplace, etc.). Once moved, it’ll fit neatly with the rest of the distribution-specific tests. If you can push a new commit for that I'll take another look to ensure linting issues are not occurring still. Evn py 3.9 |
@ryan-wolbeck Will do, I appreciate the help. I'm a bit of a noob when it comes to the formalisms of software. I also noticed some issues with my fisher information calculation so I'm going to revise that and do some additional testing. |
Added the negative binomial distribution by following the guide here: https://stanfordmlgroup.github.io/ngboost/5-dev.html