05.12.2012 Views

Student Notes To Accompany MS4214: STATISTICAL INFERENCE

Student Notes To Accompany MS4214: STATISTICAL INFERENCE

Student Notes To Accompany MS4214: STATISTICAL INFERENCE

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Differentiating again, and multiplying by −1 yields the information function<br />

Clearly ˆυ is the MLE since<br />

Define<br />

I(υ) = − n 1<br />

+<br />

2υ2 υ3 n�<br />

(xi − µ) 2 .<br />

i=1<br />

I(ˆυ) = n<br />

> 0.<br />

2υ2 Zi = (Xi − µ) 2 / √ υ,<br />

so that Zi ∼ N(0, 1). From the appendix on probability<br />

n�<br />

Z 2 i ∼ χ 2 n,<br />

implying E[ � Z 2 i ] = n, and Var[ � Z 2 i ] = 2n. The MLE<br />

Then<br />

and<br />

Finally,<br />

Var[ˆυ] =<br />

i=1<br />

ˆυ = (υ/n)<br />

E[ˆυ] = E<br />

�<br />

υ<br />

n<br />

n�<br />

i=1<br />

n�<br />

i=1<br />

Z 2 i<br />

Z 2 i .<br />

�<br />

= υ,<br />

�<br />

�<br />

υ<br />

� n�<br />

2<br />

Var Z<br />

n<br />

i=1<br />

2 �<br />

i = 2υ2<br />

n .<br />

E [I(υ)] = − n 1<br />

+<br />

2υ2 υ3 = − n nυ<br />

+<br />

2υ2 υ3 = n<br />

.<br />

2υ2 n�<br />

E � (xi − µ) 2�<br />

Hence the CRLB = 2υ 2 /n, and so ˆυ has efficiency 1. �<br />

Our treatment of the two parameters of the Gaussian distribution in the last ex-<br />

ample was to (i) fix the variance and estimate the mean using maximum likelihood;<br />

and then (ii) fix the mean and estimate the variance using maximum likelihood. In<br />

practice we would like to consider the simultaneous estimation of these parameters. In<br />

the next section of these notes we extend MLE to multiple parameter estimation.<br />

27<br />

i=1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!