> Dear Professor,
>
> In your example for the least squared error estimation (note number 13) you
> do an example. In it you let Q be a matrix of 1s. Why is this?
>
 
Answer:

mx_1 + c  approximates y_1
mx_2 + c  approximates y_2
mx_3 + c  approximates y_3 ...
 
Let A =
[x_1  1]
[x_2  1]
[x_3  1] ...
 
and Y =
[ y_1 ]
[ y_2 ]
[ y_3 ] ...
 
and Z=
[  m  ]
[  c  ]
 
Then AZ approximates Y.
 
The second column of A is Q.
------------------------------