I'm interested in examining the feasibility of assigning a difficulty rating to individual test cases. To do so I need data from a contest with partial scoring (otherwise participants won't submit unless they expect to solve all test cases). I also need to be able to determine which test cases each participant was able to solve during the contest (either the information should be directly available or the submissions and test cases should be available so I can determine it myself). Does anyone know of a platform that satisfies those constraints?
USACO is the closest I've seen so far. But full results are not available for all contests (only top participants are published for many of them) and also judging happens after the contest so it's not quite the same as immediate partial credit.
Thanks for your help!