PDF of Lecture Notes - School of Mathematical Sciences
PDF of Lecture Notes - School of Mathematical Sciences
PDF of Lecture Notes - School of Mathematical Sciences
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
2. STATISTICAL INFERENCE<br />
(3) If t(x) is a sufficient statistic for θ, then ˆθ depends on the data only as a<br />
function <strong>of</strong> t(x).<br />
Pro<strong>of</strong>. By the Factorization Theorem, t(x) is sufficient for θ iff<br />
f(x; θ) = g(t(x); θ)h(x)<br />
=⇒ l(θ; x) = log g(t(x); θ) + log h(x)<br />
=⇒ ˆθ maximizes l(θ; x) ⇔ ˆθ maximizes log g(t(x); θ)<br />
=⇒ ˆθ is a function <strong>of</strong> t(x).<br />
Example<br />
Suppose X 1 , X 2 , . . . , X n are i.i.d. Exp(λ); then<br />
f X (x; λ) =<br />
n∏<br />
i=1<br />
λe −λx i<br />
= λ n e −λ P n<br />
i=1 x i<br />
= λ n e −λn¯x .<br />
By the Factorization Theorem, ¯x is sufficient for λ. To get the MLE,<br />
U(λ, x) = ∂l<br />
∂λ = ∂ (n log λ − nλ¯x)<br />
∂λ<br />
= n λ − n¯x;<br />
∂l<br />
∂λ = 0 =⇒ 1 λ = ¯x =⇒ ˆλ = 1¯x .<br />
Note: as proved ˆλ is a function <strong>of</strong> the sufficient statistic ¯x.<br />
Let Y 1 , Y 2 , . . . , Y n be defined by Y i = log X i .<br />
99