I have an interesting problem and some information i have found, but not much. The problem can be simplified as follows. You have N schools, and each school has exactly M kids. Every school has the very same subjects, exams and scores, e.g. L subjects... A teacher in each school ranks the kid observing the scores. The teachers in other school rank the kids in a different, but quite similar way (the school has different standards). Imagine that in some of the schools, the teachers are missing (e.g. went sick). Our task is to rank for the best kid (or best top kids) by their subject score. We know that all the test group of kids belong to the same school. is there some exemplary of ML code for similar to this problem? I have tried to search for a methodology that exploits the grouping, but nothing more than LambdaMART I have found... I received very modest results...anything that ranks/regresses/classifies in a similar problem is highly valuable for me Continue reading...
