For the college game, analytics give some good overall takeaways, but nothing specific to a player or roster.
An overall takeaway would be ranking the scoring efficiency of different shots. A few aspects have been mentioned already, but knowing how to weight an open shot off the dribble vs a contested spot up gives some idea of the other variables that can be measured.
When it comes to player/roster/lineup analysis though, I'm not sure there is a big enough sample size to ever do much. Between the low number of games in a season, high variation in opponent quality, short tenure of players in college basketball (let alone tenure at a single school), tenure of coaches, rapid skill development in a single season, the "gelling" process that teams go through early in the year... drawing a valid conclusion and then coaching part time athletes to take advantage of it will be very tough.
It's not the NBA where many players spend 8-10 years playing 70-100 games per season at a relatively static skill level against nba talent every night. You can develop a good statistical basis for a strategy AND the players are around long enough to take advantage of it.
Anything you can do at the college level won't get much past normal scouting.