Is robustfit a form of MM-estimation?
7 views (last 30 days)
My understanding is that MM estimation is identical to M estimation, but rather than use OLS to estimatite the initial regression parameter we use an M estimation.I could not find any details on the robustfit documentation, but the fitlm with 'robustOpts' documentation has the line:
'For robust fitting, fitlm uses M-estimation to formulate estimating equations and solves them using the method of Iteratively Reweighted Least Squares (IRLS).'.
This sounds very much like MM-estimation, but due to the similarity of M and MM estimation and the many robust-regression methods out there I would just like to confirm in case my understanding is worng.