Recently, I have been reading and studying about the weighted LS, generalized LS, and robust regression. I sort of understand about the theory behind it and how it overcomes the heteroscedasticity, correlated error, or outliers using different weighted or using different loss function.

Now, I am wondering how frequent you came across these model or use these model when solving the real world problem (as opposed to the assignment problem)?

How does these models stack up against the tree-based model (e.g. random forest), regularized model (e.g. LASSO), or non-parametric model (e.g. GAM) or other ML models?

  • $\begingroup$ It really depends upon your application. Are you trying to make predictions or understand how different variables are related to each other? Do you expect your variables to affect the outcome linearly, non-linearly, or interactively? $\endgroup$ Jun 4, 2020 at 2:38
  • $\begingroup$ So it comes down to interoperability vs accuracy, right? $\endgroup$
    – Phume
    Jun 4, 2020 at 16:13


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy