I don't understand a step in the proof of the Gauss-Markov theorem:
Mainly, why can we simply replace y with e, given that y is defined as:
Thanks in advance!
Just substitute the $y = W\beta + e$ into the $y-W[(W'W)^{-1}W'+C]y$ (expression ahead of the red line you draw) yield:
$e - W(W'W)^{-1}W'e - WCW\beta - WCe = e - W(W'W)^{-1}W'e - WCe$
thanks to $CW=0.$
I guess the last sign in the bracket should be $-$.
Copyright © 2021 JogjaFile Inc.
Just substitute the $y = W\beta + e$ into the $y-W[(W'W)^{-1}W'+C]y$ (expression ahead of the red line you draw) yield:
thanks to $CW=0.$
I guess the last sign in the bracket should be $-$.