I'm trying to find good-enough algorithm for ranking GitHub repo "relevance", based on minimum amount of data. I have settled on a combination of log(stars) an

I'm trying to find good-enough algorithm for ranking GitHub repo "relevance", based on minimum amount of data. I have settled on a combination of log(stars) and half-life penalty for days since last commit. It works surprisingly OK, highlighting large active repos and demoting ones that aren't actively maintained. I'm sure there are many other approaches to the same thing, curious to learn how you solved it.