Turing a Least Squares function into a Weighted Least Squares

So i originally wrote my function for a least squared function but it turns out I need to write my function to figure out the weighted least squares. From my understanding, a weighted least squares problem can be solved by the basic least squares method after we multiply both Yi and the ith row of X by wi in a typical equation. But how do I should this in a matlab function?

function [A, e] = WeightedLeastSquares(X, Y, w) size(A) rank(A) rank([A b]) X = A\b Y = norm(A*X) w = pinv(A)*b

I already have the variables A,e, X,Y,w loaded onto my current workspace, but I just need to change my function. So, if I call,

load Lab08Data.mat
I get
[A, e] = WeightedLeastSquares(X3, Y3, w)
A =
3.0573
1.9250

e =
0.0218
-0.0702
0.3235
0.0552
-0.1038
-0.0202

Any suggestions on how turn my function to a weighted least squares function?

[SB]

Hey, I'm stuck on the same question! How did you get the original least squares polynomial regression btw? I tried function [A,e] = MyPolyRegressor(x, y, n)

for i = 1:n

   b(i) = x(i)^(n-i) ;

end

A=b\y'

e=b*A-y'

But it didn't work.

[Ben]

you just have to use polyfit and it should work!

[SB]

I dont think we're allowed to use polyfit; it defeats the purpose of the problem.