In most practical cases, models are burdened with collinear and noisy features. while regularization helps avoid overfitting by effectively simplifying the model, the question of interpretation of the relative importance of predictors in the "real world" remains. How can we tackle this issue?
Asked
Active
Viewed 18 times
1
-
Possibly relevant: http://stats.stackexchange.com/questions/202277/what-are-variable-importance-rankings-useful-for – Matthew Drury Dec 14 '16 at 21:10