We're almost at real machine learning and modern data science. But to understand what makes modern data science so powerful, it is worth undrstanding the very old problems that it solves. Figuring out how to have results that generalize to conditions beyond the data at hand is the hardest part of machine learning, and something we'll spend a fair bit of time exploring in the course. Traditional statistics focused on many questions, but never really solved this core problem. The best we could do was hold out some data. As we'll see next week, that was a great idea; computer scientists saw it, and ran with it.
Please read Chapter 3: Arms Race, Going to College in Weapons of Math Destruction.