I want to fit a robust regression method to my data because there are some outliers that might influence the estimates too much. Now my question: Are criterions like AIC or BIC still useful for robust regression models in order to compare models with different variables (variable selection)?

  • 1
    $\begingroup$ I don't know robust regression, but how do you want to use the AIC or BIC? If it is for comparing the model with and without the outliers, usually you can't use those criterions for that. Models compared need to have the same set of observations. $\endgroup$ – Emilie Nov 6 '15 at 14:23
  • $\begingroup$ I want to compare robust regression models with the same set of observations but with different predictor variables. Hence, the BIC would be my criterion to select variables $\endgroup$ – R_FF92 Nov 10 '15 at 6:40
  • $\begingroup$ You are probably looking for this. $\endgroup$ – tchakravarty Nov 10 '15 at 7:24
  • $\begingroup$ Yes I think this will help me. You know whether this robus AIC procedure is in R? $\endgroup$ – R_FF92 Nov 10 '15 at 7:35
  • $\begingroup$ Not really; you will have to look through the robust regression packages (cran.r-project.org/web/views/Robust.html). But this would not be hard to code up at all. $\endgroup$ – tchakravarty Nov 10 '15 at 7:37

Your Answer

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

Browse other questions tagged or ask your own question.