This is probably the most performant solution here. In production, we always do overnight snapshots of highly computationally intensive data so that it can just be retrieved quickly instead of forcing the end-user to wait for the data to be computed on the fly. So while it may not be the most elegant solution out there, it's probably the most appropriate in a corporate setting where a certain set of data won't change much (or at all). Good job!

It would be awesome if they just made the median function part of the language.