0 ratings0% found this document useful (0 votes) 47 views6 pagesLecture Notes - Week 12
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content,
claim it here.
Available Formats
Download as PDF or read online on Scribd
The Method &f Maximum Likelihood
Wi'dely used method do obfain an estimator poraa
vnkaown povametir 2] popiladion with o pita obied
vs called fhe method of moximum Ubth’Lood .
Lf 2, %,)--- 2%, be an observed panclem sample fom
alist, with peramehe @. Llebshood forebion ab
Neen
A ee i fen)
= IT {aed
Max, Ukedfood estimador (MLE) of 6,6 maxilal!
ether Mikelnood fuachlon, £y 6) or lop thmi'e
tkdihood function &,(0) = ba Ly lo).
Nole that bikebrhood funchon rs oedvally forn F
probability fl obsecviag He dala set (sample) we
Aase.
Ln dserele case
PUK = X= % 5 --- n= Xa) = POX 2%) POL = 35) PUM)
(hecovee X's ave sndependent )
= 29) fla 2) [8 ) £09)
where Jere) afte) = PUX= %)
La cond. Cuse
PUK 22) XarBa yo he = /th20)- 8) -- fare)
{4 ® 1s fonchion elve d/ ({ for tach 3%".
> L00= fT f08)
C##) avetiant peperly of MLE’
MLE's are savariant onder the transformations
pocameler, Suppose that 6 ts a MLE f & ond
Me hte) Is ferehan of & thea MLE of tb
Pe A(@)
xan ple.
Zt XX 2% be a random sample of ty Xa expla)
Find a)MLE& of X, £) MLE A = ECX)-
HA!
{aprAyeAe " x>0
a) ZA = TP Ae. Then Kyla\edn byl)
oe fA i a -AK!
2 &M= ba [PAE Ja OE") |
a =F lod Ax) aalad — DEK
=> 4- Em © 7
2 n-rZx=0 MLE Ad,
Set - Lb
2x; x
4) He ER) =f 44 fhe saveerant propels a 4) MLE;
dhe 9
od
MLE of p, bo Ee f=
(78) Zxample
The atetel ies sold mn a olay ba Poison bv
with parameder (pitan) , 4. A cettash car dealer
sofd no car rn JO aes, at bast one car th each 4
He rtmaing 20 days of a month. Fibd the aaa
Uhelthood estimate of #.
oH xe
(AW) = eu Xie OND yee
a
POKi20) = ety ze (poh, dev each +f 10 ays)
Hin PUXIS1) -100/-E") p o0é" = oO
=) -107108 "4208" = 0
=> 30E"%-/0
cise or -hz-&n3
/
=
=> ML, Eshmade of Hy = On = 109
CFF) Latevel Exkinahion Confidence Daher ved's
For a forametec 0, with an esh'imador ©
welll obtern lower confidence. Lint, 9 (6) and
Upper confr'dence. bint, og B), 20 that with
Jak confidence uel , pre valve of Povamefer O-
will be ro the /aperval
C4@) » pF) ) .
TAi's pakervek 1's called yo9 1/-k) Yo conf s dence
thfervad CCL) f'
preb. Like 0,90, 0,95 er 0,99, ede.
Leye-sampe er for fl
For lay. samples, 44 CLT we know Shot
estimator 4 Uy samyee mean,X 7s app. non ally
disbebeded. Then 7
seas ne std. normed .
vn
Zt folk confr'dence Level. Well obtain
fo0Cl-“) 42 CL for y :
ee Ae
POR EEE Agee Lhe \ os
PC -Be L id a
~ 2g 7
> where /—
ib x Adph
PR
= 2
aaa
we'll be ey opecabions i'n efeality und] we have
the mide le
(4 wh the mid (0)
WE oy ete
PC 8g KK MS 2g Baw
a i
PC-X 24 <-H4-K% ee, )onw
EVA
x. = 57 \
OOK + Bg Fe > US Fee ) ape
=D /000/1-L) Yo CFI fo
(K-Fe2e , Fp Ea) or KES Oy
4] 1s enkaown le std, cle, 5 compuled — |
fom Sader, (7 Jotpe an dle MYfferen ce. bedueca
<= and = can be neylecled, the cL for 4 |
will be KAS Be |
Tne |
Queshon_
Suppose thet with neso observahone 95 Yo
er (or Aids foved fo be
C389, 412) .
a) Ts sample mean, ® rn this saberval
4) Ls population mean Hy 1 this tnberved P
Exrplaro-
a) Sine Mbhrva/ 1's found using XtS2,
in 2
samp. mean XK /'s alwag s to the sakerval, fooad
Db i's midpoint of the CL.
Cb b) Yes or Wo (er cannot- 4Y)
We ae 9540 confy'dent Phad the trve meen of
He papolahon sin dhe thdervel (3.89, 4,12)
LY wt continve samplihg and obfuh 95 Jo
CL from each samp of Samet ste (Nz 50) welll
obtenn AMferen t CLS Checovee Kands wif
Be Mifferen F each Sime, tun thouph we know
a, KX willl chanpe \, then apprex!matelg ISS
eh thece tnbervels willl shehude Prue mean ut.
and OPP. se of Phen wlll nef pckude the.
Jive valve af dhe meen Y.
Nole that CI's are random, heceavoe Upper
and flower baits eure frchons of sandanry
Samp , bet meen of oo popelablon i's a heel,
constant. In fhe digere above we assumed thof
ped
(82)
