Scribd
Upload
175 views15 pages

Lecture 2727K19EN

mth305mth305

Uploaded by

Yeementh Virutkar
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
175 views15 pages

Lecture 2727K19EN

mth305mth305

Uploaded by

Yeementh Virutkar
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
You are on page 1/ 15
Properties of Maximum Likelihood Estimators. (™ Le E d _ Se 2 (0 The frst and second order derivatives, vit. 2708 ang POEL exist St and second order derivatives. vt and are continuous functio i R (including the true value 8p of the Parameter) for almost all x. For every 0 in R < Fx) |S.e2| log L exists such that oe 2 demea| i) saa - log L| < M(x) _ Ea an pry where E(M(x)] < K, a positive quantity. € (d= iii) For every 0inR, | Mon ae Snr) J am {sen is finite and non-zero. Atv) The range of integration is independent of 9. But if the range of integration depends on 6, then f(x, 6) vanishes at the extremes depending on 0, This assumption is to make the differentiation under the integral sign valid, a (Cramer-Rao Theorem). “With probability approaching unity as a> =, the likelihood equation & log L = 0, has a solution which converges in probability to the true value @)”. In other words M.L.E.'s are e ther words M.L.E.'s ar consistent. ~— Remark. face's are always consistent estimators but need not be unbiased. |For example in sampling from N (jt, 62) population, [c.f. Example 15-31], MLE(u1) = x (sample mean), which is both unbiased and consistent estimetor of pt. MLE(o?) = s? (sample variance), which is consistent but not unbiased estimator of 62, p{\t eleehor me xr (Hazoor Bazar's Theorem). Any consistent solution of the likelihood equation provides a maximum of the likelihood with probability tending to unity as the sample size (n) tends to infinity. mos p = vw (Asymptotic Normality of MLE’s).4 consistent solution of the likelihood équation is asymptotically normally distributed about the Q. Thus, 6 is asymptotically N (0 ito) n— oo, Remark, Wann of ali is given by v@) =Ta 1 TCE] Zo we 7 Af MLE. exists, it is the most efficient in the class of such estimators. —_____ STheorem i suficen extinaor axes is a fction of the Maximum Likelihood Estimator. —Proof. If = 1G, 2, ra) is a sufficient estimator of 9, then Likelihood Function can be written as (cf. ‘Theorem 15-7) L = g(t, 8) h(a, x2 X3, «--s Xal where g(t, @) is the density function of t and h(x, x2, function of the sample, given t, Den ate al h be jo" y he ba weey7? X11) is the density log L = log a(t, 0) +10g ACs, %2,--¥4!0) inn] a Differentiating wert. 0, MAL log L ! DRL 3 tog g(r'®) = v(t, 0), Cay) which is a function of t and @ only. MLE. is given by d log L =0 = ywlt6)=0 20 (8 =n(@ = Some function of sufficient statistic. LP = (@) = Some function of M.L.E. —_ ae If for a given population with p.df. fix, 6), an MVB estimator T exists for 9, then the likelihood equation wifl have a solution eq to the estimator T. Proof. Since T is an MVB estimator of 0, we have (c.f. (15-40)], a T-0 F) log L = x) = (T -@) A(@) MLE for 6 is the solution of the likelihood equation Diogh=0 = 6-7 Hplesl=0 => = as required. fi — (Invariance Property of MLE). If T is the MLE of Band y(6) is one to one function of 6, then y{T) is the MLE of y(). Te yn 2 Pw Gal — Find the maximum likelihood estimate for the Parameter 2 of @ Poisson distribution on the basis of a sample of size n. A (yen Potrsmma Jone etx fern ar Lsethant — fonaion | Le mite) fst Prat ty a mcopnen — poe 5 oun Sar) ooo 4 cua) ——® > %& oN Ae ->d Xy Lze A @ a e A —<$— . ad ——_—— HN AI An Xn » -»V" my Me x Gl _. Kal =x" WitAga- # Am —— Ay) %1 -... Kal ty [er om) Al Al... An | we aan - tae n+ Lun) om!) = -AN dye + % | a e C Sey a) A " : ot an!) My —dX\n L (Gras peck Ay ) & x mal) | _ assy (a1 x at ac xe Cb b An) aia fg d » = 4a Cay Alo: My) - -m&) + (mbar day Lao > vy 2 ont eS da 2 Like Ja bay <= stn on goo + x + a J + x x 3 i ° -1 mia eb Em 7 OT >» yr Xa be # hm n= n

You might also like