Professional Documents
Culture Documents
GLM Lec3
GLM Lec3
GLM Lec3
Goel
Broad Outline
Training Sample Model: Given n observations, [ [(Yi , x′i ), x′i = ( xi1 , , xir )] ,
i = 1, 2, n, the sample model can be expressed as
=Yi µ ( xi1 , xi 2 , =
, xir ) + ε i , i 1, 2, , n, (1)
where, ε i , i = 1, 2,, n , denote the noise (random errors), each with mean
zero and variance σ 2 .
∑ β j xij + ε i , E[ε i ] =
Yi =
j =1
0, Var[ε i ] =
σ 2 , Cov(ε i , ε k ) =
0, i ≠ k . (2)
p
] µ=
E[Yi | x i= i ∑β x .
j =1
j ij (3)
Vector/matrix notation for the response, predictor variables, error terms and
the unknown coefficients:
Y1 β1 ε1 x1.′ x1 j
Y β ε x′ x
=Yn×1 2
=
, β p×1 =
2
,ε 2
=
,and X n× p 2.
let x. j ,
. Also,=
2j
′
Yn β p ε n xn . xnj
denote the jth column of X , i.e., X n× p = [x.1 , x.2 , , x. p ].
Given the response vector Y, and the design matrix X , the sample GLM can
be written as
Xβ + ε , E[ε ] =
Y= 0, Cov[ε ] =
((cov(ε i , ε j ))) =
(( E (ε iε j ))) =
E[εε ′] =
σ 2I. (4)
Thus, E[ Y]= μ= Xβ , Cov ( Y)= E[( Y - Xβ )( Y - Xβ )′]= E[εε ′]= σ 2I. (5)
Ordinary Least Square (OLS): For an estimate β of β , corresponding
n p n
∑ (Yi − ∑ β j xij )2 =
Residual Sum of Squares: l2 ( β ) =
=i 1 =j 1
∑ ei2 (β ) =
=i 1
e′( β )e( β ).
In matrix notation,
• min
e′( β )e( β )= min(
Y − Xβ )′(Y − Xβ )= e′( βˆ )e( βˆ ). (6)
β ∈R p
β ∈R p
• E β=
0
( X′X) − X′E[=
Y] ( X′X) − ( X′X=
) β Hβ ≠ β , so some components of
β do not possess unbiased estimators.
Why?
Optimal with respect to what criteria? May need to add an
additional criterion to OLS to get an unique solution, e.g.,
Even though in non-full rank there are infinitely many solutions (in β ) to the
normal equations case, the projection µˆ = PY (also called Yˆ ) is unique, i.e., the
matrix P does not change with the choice of a generalized inverse of X′X .
Pu = X( X′X) − X=
′u X( X′X) − X′Xb
= Xb
= u.
• If there does not exist any unbiased estimator of the linear function
c′β , it is said to be non-estimable.
• c ∈ C=
( X′) ⇔ c X′l for some l ∈ n .
Therefore, for any OLS β 0 , c=
′β 0 l′X ( X ′X ) − X
= PY l′µˆ .
′Y l′=