This is a super cool project! It would be especially interesting to use on pro-level games, maybe you could even sell services to pros who want to get better. Like being able to say "your baneling run-bys are most cost-effective with 7 banelings around the ten minute mark" etc. With a relatively larger dataset you could come up with some real interesting statistics on individual players performance, use of certain un…
If it's quite good it could be useful as a tool for people trying to get better (although various tooling and custom maps have been around for these types of purposes since the game came out and most people even pros seem to just grind ladder for practice).
I was also hoping it could be used to analyze a replay and indicate how "back and forth" the game was, or if it was more one-sided, etc. Or if the top 100 pros were part of the featurization, then you could potentially determine how the odds change given a certain game state depending on who the pro is.