With the rise of complex machine learning models like neural networks and random forests, are linear models still relevant? In what scenarios do you find linear models to be effective or preferable? Would love to hear your thoughts and experiences.
Interpretation is really important. A linear model is far simpler to describe than a neural net, which is important if you want your models to clear regulatory review and receive the go-ahead for commercial use. It’s dead on arrival if these models cannot be adequately explained.