This looks interesting, I bookmarked it.
My biggest blocker is the "statistics" part of M/L, knowing what algorithms to choose for various cases.
My biggest blocker is the "statistics" part of M/L, knowing what algorithms to choose for various cases.