ML blog - polynomial features

Follow the full discussion on Reddit.
In ML beginner courses, when teaching linear regression for curve fitting, they tell us that high degree polynomials are a big no no! We're told they oscillate and overfit, and can't be controlled with regularization. Well, I hope to convince you that, to some degree, it's a myth. Here is a first post in a series: https://alexshtf.github.io/2024/01/21/Bernstein.html