Nadaraya-Watson vs Lagrange Interpolation

Ok, I don't get it, what do Nadaraya-Watson estimators have over Lagrange Interpolation?
Since the choice of Kernel doesn't matter much, you can just set k(u)=u, right..?
What am I not getting here?
(Kernel methods involve Convolution Smoothing)
                - = - Vasos-Peter John Panagiotopoulos II, Columbia'81+, Bio$trategist http://ourworld.compuserve.com/homepages/vjp2/vasos.htm ---{Nothing herein constitutes advice. Everything fully disclaimed.}---
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.