Linear regression - least squares method to solve

• 1.

　　　　Overview:

　　　　　　Linear regression is the use of mathematical statistics, regression analysis, to determine a statistical quantitative relationship between two or more interdependent variables analysis methods, the use of very extensive.

　　　　The expression of the form y = w'x + e.

　　　　demo:

　　　　　　For a set of data {(100,20), (160, 30),), (60, 15) ............}, assuming x, satisfy some linear relationship between y. Given objective function y = ax + B process, even when solving the linear regression procedure a, b, and

　　　　One solution is the least squares.

2.
the y-w'x + = e, because e is composed of various uncertain factors, it can be introduced by the central theorem of the normal distribution with mean zero.

 

 Release parameter vector is: (XTX) 'XTy