Programming WLS function excel vba - vba

Question

I am attempting to replicate a function for WLS (Weighted Least Squares) that I have found in a textbook in excel. There is a value error coming up and I think that I am doing something wrong in using the function.
The following is the VBA code for a supporting function Diag(w) and the function itself WLSregress():
Function Diag(W) As Variant
Dim n, i, j, k As Integer
Dim temp As Variant
n = W.Count
ReDim temp(n, n)
For i = 1 To n
For j = 1 To n
If j = i Then temp(i, j) = W(i) Else temp(i, j) = 0
Next j
Next i
Diag = temp
End Function
Function WLSregress(y As Variant, X As Variant, W As Variant) As Variant
Wmat = Diag(W)
n = W.Count
Dim Xtrans, Xw, XwX, XwXinv, Xwy As Variant
Dim m1, m2, m3, m4 As Variant
Dim output() As Variant
Xtrans = Application.Tranpose(X)
Xw = Application.MMult(Xtrans, Wmat)
XwX = Application.MMult(Xw, X)
XwXinv = Application.MInverse(XwX)
Xwy = Application.MMult(Xw, y)
b = Application.MMult(XwXinv, Xwy)
k = Application.Count(b)
ReDim output(k) As Variant
For bcnt = 1 To k
output(bcnt) = b(bcnt, 1)
Next bcnt
WLSregress = Application.Transpose(output)
End Function
This function should return the WLS estimator for the explanatory variables of equation being estimated. I understand the code leading up to the k = Application.Count(b) line but not too sure how the output bit is working.
If anyone could help me figure out why this isn't working I would be very grateful.
The following is an example image of the function trying to work.

Answer 1

By default, Excel will start sizing its arrays with 0 if you don't indicate otherwise. For example,
Redim arr(2,2)
will actually give you a 3 x 3 array
0 1 2
0 blank | blank | blank
1 blank | blank | blank
2 blank | blank | blank
Because of this, when you have ReDim temp(n, n), you're actually creating an array with one more row and column than you actually want. In your example, you would expect the Dialog for A3:18 to be a 16 x 16 dialog, but it will actually create a 17 x 17 dialog, throwing off your matrix multiplications (i.e. Application.MMult)
Replace this line
ReDim temp(n, n)
With this line
ReDim temp(1 to n, 1 to n)
And you should now get results returned. Up to you to determine whether the result is accurate or not.