Hacker News new | past | comments | ask | show | jobs | submit login

Explanation of results for non-ML folks (results on the default supabase repo shown on the homepage):

Codeball's precision is 0.99. It simply means that 99% PRs that were predicted approvable by Codeball were actually approved. In layman, if Codeball says that a PR is approvable, you can be 99% sure that it is.

But recall is 48%, meaning that only 48% of actually approved PRs were predicted to be approvable. So Codeball incorrectly flagged 52% of the approvable PRs to be un-approvable, just to be safe.

So Codeball is like a strict bartender who only serves you when they are absolutely sure you're old enough. You may still be overage but Codeball's not serving you.




A LOT of ML applications should be exactly like this.

I want systems with low recall that "flag" things but ultra ultra high precision. Many times, we get exactly the opposite - which is far worse!


Here's a visual explainer on Precision vs Recall (in the context of ML algorithms):

https://mlu-explain.github.io/precision-recall/


That’s still super useful.

I’m assuming most PR’s are approvable. If that’s the case then this should cut down on time spent doing reviews by a lot.


So basically, very few false positive but lots of false negative is the tradeoff made by Codeball?




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: