When studying non-parametric regression, I've been told that local linear fitting is often better than local constant fitting at the job of estimating regression functions because local linear fitting is, unlike local constant fitting, "design adaptive". Despite my best attempts at interrogating Google Scholar, I can't seem to find a definition of this term. I suspect that I'm missing something obvious. For example, is this just a bloated way to say that local linear fitting is more flexible than local constant fitting?
$\begingroup$
$\endgroup$
Add a comment
|