data y=xb+E with noise E~N(0,sigma)
the least square estimator b' satisfy E||xb-xb'||^2=(number of parameters in
b)X(variance of noise)
不是很理解这个number of parameters.
w*******9 发帖数: 1433
2
Beta is multiple dimensional. The result is easy to prove by considering the
trace of Xvar{hat-beta}X'
N******n 发帖数: 3003
3
the
~~~~~~~~~~~~~~~
看了半天，好像还是不清楚？3x
【在 w*******9 的大作中提到】 : Beta is multiple dimensional. The result is easy to prove by considering the : trace of Xvar{hat-beta}X'
