This preview shows pages 1–6. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: begin/i Hon ( ’ {Tami/“.54
LM Z {a & (JRHSH M M or— aw
{ﬁe/[aging fa {am/ﬁg (m p P lee®§ I M 44% ‘PA 6 66/0 0; 611414
3e} ® a FﬁZH/Eaé M ﬁﬁmﬁ PMM %ﬂ(‘2}9§ : 66 69}
1s hie mgr ﬁne _M0tWLC(‘€/\ 9&4 H3
m Eefuézﬂ o M an (gee?)
MAN?“ [**8 Pa[%(2\: 0] :1 ﬁx a“ 966). EKQMPLL Z 51 é‘tnowvia/Q (L/{RML Pawn
«3393 = (‘28ez(‘(~e>“‘z 1 TM WW A€\2~6 10<8<1§ ‘
WNLg‘ﬂrg This. (rt/13641433: _€/§WAHons Z=Oﬂ1wUMl Nakga : Hm m ﬂaw WWMJ Maw {.7me § 0: Eéwzx] = :uzzx—ge'G J e>o ivx EM ‘ 0 E V
MW wmiﬂw 0 == MOJ WW) = Nahﬁz: Hm we (1m iné'mfefg ma ' MW }
m mg (m {N ew& Wm o€€> m we
gawk/5 Mal/lg Mae/64 14(0)) 74(7)) ,, a a _. ,6»; 3
LC 5:chst =0) gear am 156: (aw
{AW (:90?) = 0 {3w almosrL all We (ch (a?) ,
This; MQWIS: 14: Z is a lww/iwm mmﬂom
mﬁa/{lﬁ‘é MM (la/w 56W 6% am! 5
(ca/m? Z W a pﬁhﬁ ) HLML H423 : 0] =10 (ME/Laaapaf)
P[Q[Z)=O] =15 Cam {ék Owl £3:ch gﬁl c @Qw¥£¢¢_9€®‘
v ‘ 4 4 "I I M‘ 03f PWDCS E x 3 gay 93 g Q 6 ﬂ EW::;_MAnH;:W f" 7 s , I 2 s, e ie “A j m “ m __ ___.;_L<J:L ‘5“ 4er 61% 5k MR matamwa , _ Z m m Em eA ,wﬁm.4zbr_g%ﬁ E
cm > ém £0 659 a” hi '
ems? v wt Magi M g :vakckmw: of. . . y' ’ I K? but I” A
I” v u~
I 0' ’i // 7.5 Exponential families of Disfhbabons Consider a family f($;6! : 6 E Q) of probability density or mass functions, where Q is the interval set 9 : {6 : 7 < 6 < 6}, where 'y and 6 are known constants
(they may be ioo), and where ﬁx; 9) Z { eXPlP(9lK($l + 5(a) + q(6)] a: e 5 O elsewhere, (751) where S is the support of X. In this section we will be concerned with a particular
class of the family called the regular exponential class. Deﬁnition 7.5.1 (Regular Exponential Class). A pdf of the form (7.5.1) is
said to be a member of @ regular exponential " ; of probability density or mass functions if 1. S, the support of X, does not depend upon 6, 2. p(6) is a nontrivial continuous function of 6 E Q, 5’. Finally, (a) ifX is a continuous random variable then each of K’(ac) i 0 and S($)
is a continuous function of a: E 8, (b) if X is a discrete random variable then K is a nontrivial function of
a: E S. For example, each member of the family {f(:c;6) : 0 < 6 < 00}, where f(:c;6) is N (0, 6), represents a regular case of the exponential class of the continuous type
because 1 —a:
f($;6) 2 Tree 2/29 1 .
exp (—Eémz — logv27r6>, —oo < a: < oo. II On the other hand, consider the uniform density function given by ﬂag) 2 { exp{——log 6} a: 6 (0,6) 0 elsewhere. This can be written in the form (7.5.1) but the support is the interval (0,6) Which
depends on 6. Hence, the uniform family is not a regular exponential family. Consider a family { f (:13; 0) : 49 E Q} of probability density or mass functions,
Where Q is the interval set 9 = {6 : 7 < 6 < 5}, Where *y and 6 are known constants
(they may be too), and Where . _ eXp[p(0)K(w) + 3(93) + 9(9)] 03 E 5
ﬂw’g) — { 0 elsewhere, (7'5'1)
Where 8 is the support of X.
Let X 1, X 2, . . . ,Xn denote a random sample from a distribution that represents
a regular case of the exponential class. The joint pdf or pmf of X1, X2, . . . ,Xn is n exp [p(6) Z + Z + nq(6) 1 Theorem 7.5.2. Let f(a:;6), *y < 9 < 6, be a pdf or pmf of a random variable X
whose distribution is a regular case of the exponential class. Then if X1,X2, . . . ,Xn (where n is a ﬁxed positive integer) is a random sample from the distribution of X ,
n _
the statistic Y1 : ZK is a suﬁicient statistic for 6’ and the family {fy1(y1; 0) : 1
7 < 0 < 6} of probability density functions of Y1 is complete. That is, Y1 is a
complete sufﬁcient statistic for 6. ...
View
Full
Document
This note was uploaded on 06/11/2010 for the course STAT 410 taught by Professor Alexeistepanov during the Spring '08 term at University of Illinois at Urbana–Champaign.
 Spring '08
 AlexeiStepanov
 Statistics, Probability

Click to edit the document details