Model Selection for Optimal Prediction in Statistical Machine Learning
Ernest Fokoué · Notices of the American Mathematical Society · 2020
At the heart of this article is the theoretical result known as the no free lunch theorem, which reveals, both implicitly and explicitly, that the theoretical bounds studied extensively by experts do not help much when it comes to practically selecting the optimal predictive model. Optimal predictive modelling is both a science and an art, requiring both mathematical and statistical rigor along with practical computational common sense.