Least Square Problems⚓︎
Given
where
We will introduce three different methods to solve the Least Square problem.
Normal Equation⚓︎
Traditional Method⚓︎
We solve
The Least Square solution is:
Question
: How about
Comment
: This method is not recommended is
Note:
Usually in practice, the number of matrix rows are way larger than that of columns.
Two Recommended Methods⚓︎
QR Factorization⚓︎
Let
SVD Method (Recommended)⚓︎
Assume
Define
We get the formula:
However,
The pseudo inverse
Then
We can rewrite another formula of the result:
where
Comments
: If
We can compute the approximate value ("the best approximation of
This is the idea behind Principal Component Analysis (where to cut off?).