View Full Version : Please help with simple linear regression!
02-03-2010, 08:22 PM
Heres the problem, i had to post a screen shot of it because of all the algebraic symbols..
And the Linear Regression model given in 2.4 is y = B1 + B2x + e
(those B's are beta's, just like in the link above)
02-04-2010, 02:50 AM
you can prove this geometrically. Let C(X) be the column space of the matrix [1,X] consisting of 2 columns (1st column is n-dimensional vector where all the entries are 1, and 2nd column is your n-dimensional vector X, so the matrix is of size n by 2) now yhat - ybar belongs to C(X) and ehat is orthogonal to C(X) so the sum in the question is nothing but the dot product between them so it equals 0 since they're orthogonal. Hope this helps!
02-04-2010, 12:57 PM
holy cow thats confusing, anyway you can show it by hand?
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.