(*)

)
/

Maximum (Method

) Ordinary Least Squares Method

) Likelihood
Non-Informative Prior Density Function
.

.
/ /

.(
"

________________________________

(*)

"
/ /

(
(

)
)

.
.

.
Regression Analysis

.
.Linear Regression and Non-Linear Regression
.

...

(k)

(Y)

( X 1 , X 2 , ... , X k )

Yi = 0 + 1 X i1 + 2 X i 2 + ... + k X i k + ui

...(1)

(n)

(n)
:
Y1 1 X 11
Y 1 X
21
2
. .
.
=
.
. .
. .
.

Y0 1 X n1

X 12

. . .

X 22

. . .

. . .

. . .

. . .

X n2 . . .

U1
U
2
.
+
.
.

U n

. ( 2 )

Y = X + U
[ n x (k + 1)]
([ k + 1 1])

X 1k 0
X 2 k 1
. .

. .
. .

X nk k

...(3)

.(n 1)

:Y
:X
:

.(n x 1)

: U

Y1 0 + 1 X11 + 2 X12 +...+ k X1k +U1

Y + X + X +...+ k X + U
2k
2
2 0 1 21 2 22

.
.
.
.
=
...(4)

.
.
.
.

.
.
.
.

Yn 0 + 1 X n1 + 2 X n2 +...+ k X nk + Un
(X)

( X)
:

[(k + 1) x (k + 1)]

. . . X1k 1
1 . . . 1
. . . X 2k X11 X21 . . . Xn1
. . . . .
.
.
.

. . . . .
.
.
. . . . .
.
.

. . . Xnk X1K X2K . . . XnK

1 X11
1 X
21

. .
X X =
. .
. .

1 Xn1

i =1

. X i1 X ik
i =1

n
.
X 2 ik

i =1
n

X i1

X 2 i1

X i1

i =1

i =1

.
.
.
n

i =1

i =1
n

X1k

.
.
.
n

X ik X i1 .

i =1

X ik

... ( 5 )

Ordinary Least Squares Method

( )
(k)

Yi =

i1

i2

:
U i = Yi 0 1 X i

... ( 7 )

(n)
( 1 )

( 0 )

+ ... + k X

ik

+ U

... ( 6 )

...

:
= ( X X )

... ( 8 )

( )

:
Var

( X Y )

) 1 2 ] ... ( 9 )

( ) = [ ( X X

cov

(Maximum Likelihood Method)

(Anderon and Bancroft , 1952)

( )

(n)

( X 1 , X 2 , ... , X n )

f ( X , )

(L)

(Harrison , and Mike.,1989)

:

L ( ) = f ( X 1 , X 2 , ..., X n )
:
n

L ( ) =

i=1

f ( X i, )

L ( ) = f ( X 1 , ) . f ( X 2 , ) . . . f ( X n , ) . . . . ( 10 )

:
.
.
ln L

. ( )
.
( )
= 0

.
.
.
.

(n)

( Y1 , Y2 ,..., Yn )
( 0 , 1 , u2 )

( Yi 0 X i ) 2
n
1

Pr ( Y1 , Y 2 , ... , Y n 0 , 1 , 2 ) = ( 2 u2 ) 2 exp i =1

2
u

. . . ( 11 )

:
n

n
n
2

X
Y

X i
i
i
n

0 = i=1

i=1
n

n X
i=1

X i Yi

i=1

2
i

i=1

( X i )

. . . ( 12 )

i=1

or 0 = y + 1 x
n

1 =

i =1

Yi

i =1
n

i=1

n X i2 ( X i ) 2
i =1

. . . ( 13 )

i =1

:
( ML ) = ( OLS

n X i Y i Xi

(M L)
)

. . . . ( 14 )

OLS)

...

( ML ) = ( X X ) 1 X Y

. . . ( 15 )

(ML)
.
.
( = 1 , 2 , ... , n )
.(Prior p.d.f)
.
( )

. f (Y )
(Loss function)

(Y)

(Posterior p.d.f)

Prior p.d.f

Initial Info.

Posterior p.d.f

Bays theo.
New data

Likelihood F.

:
.(
.(

)
)
.

.
.
.

.
.

(Noninformative prior distribution

function)

Jeffry

( )

, )
(
:

( )

( 0 , )

.
( Yi )

Likelihood function
n

L ( 0 , 1 , u2 ) =

f ( Y i / X i , 0 , 1 , u2 )

i=1

=
(

2 )

exp [

) n un

( Yi 0 1 X i ) 2 ]

. . . (16)

2 u2 i = 1

:
L

1
un

exp [

2
( Yi 0 1 X i ) ]

2 u2 i =1

. . . (17)

( 0 , 1 , u2 )
:

F ( ) = [I x ( ) ]

Jeffry

1/ 2

f ( 0 ) cons tan t

< 0 <

. . . ( 18 )

...

f ( 1 ) cons tan t
f (

1
u

< 1 <

0 <

2
u

<

... ( 19 )

Improper

.Proper
(17)

(19)

f ( 0 , 1 ,u Y )
< 0 <

Jeffry

1
un+1
,

exp[

1 n

(Yi 0 1 Xi)2 ] ...(20)

2u i=1

< 1 <

0 < u2 <

( u )

( 0 , 1 )

( u )

(20)
n

f ( 0 , 1 Y ) [ V S e2 + n ( 0 0 ) 2 + ( 1 1 ) 2 X i2 +
i=1

n
n

2 ( 0 0 )( 1 1 ) X i ] 2
i=1

. . . ( 21 )

[ V S e2 + ( ) X X ( ) ] n / 2 K ( 22 )

< 0 < , < 1 <

(t Bivariate)

(t)

(22)

( 1 )
( 0 )

V +1
2

n ( X i X )2

2
i
=
1
f ( 0 Y ) V +
( 0 0 )
n

Se2 X i2

i =1

:
(Student-t)
n

L ( 23 )

(23)

2
n
2

n
X
X
(
)

( 0 0 ) tv
i =1 n
2

S2
e Xi

i =1
. (v)
(t)
( 0 )

( 0 )

L ( 24 )

( tv )

(24)
( 0 )

(V)

:
n

i =1
n

i =1

X i2 Yi ( X i ) ( X i Yi )

0 = i =1

i =1

X i2

i =1

( Xi )

. . . ( 25 )

i =1

( 1 )

(22)

( 1)

( 0 )

...

n (X i X )2

( 1 1 ) 2
f ( 1 Y ) V + i =1
n

u2 X i2

i =1

V +1
2

L ( 26 )

(Student-t )

n
2
(X i X )

i =1
S e2

(V)

(26)

1
2

( 1 1 ) tv

L ( 27 )

( 1 )

(t )

(27)
( 1 )

( 1 )

1 =

i =1
n

i =1
n

i =1

n X iYi X i Yi
n

i =1

X i2 (

i =1

i )

. . . ( 28 )
2

.
)

(
.

(150)

(Y)

yi = 0 + 1 X i1 + 2 X i 2 + 3 X i 3 + 4 X i 4 + 5 X i5 + U i

VAR
(OLS)
VAR
(ML)
VAR
(BAYES)

872.038

48.068

72.34

0.754

0.156

86.225

872.038

48.068

72.34

0.754

0.156

86.225

592.559

45.451

69.582

0.720

0.1502

80.455

(ML)

(OLS)
:

eff ( i ) =

Var ( i ) in ML
Var ( i ) in OLS

(ML)
(OLS)

.
.

.
.
.

...

.
.
.
General Least Squares
Square
Minimum Variance

Moment Method

Method
Minimum -Chi Method
Method

.
.
"

"

.
.

"

"

.
"
.

.
"

1. Anderon and Bancroft,Statistical Theory in Research Mcgraw- Hill Book Company

Newyork, 1952 .
2. Harrison, and Mike Baysian Forecasting and Dynamic Models Springerverlag , New
york, 1989 .

Some Classical Estimation Baysian Methods and Estimation For

Parameters of General Linear Regression Model
(A Comparative Study With Medical Application)
ABSTRACT
In this study a comparison between classical estimation and Baysian estimation have
been done two ways of the classical method have been chosen namely (Ordinary least
squares method) and (maximum likelihood method) and only one method of the Baysian
estimation has been chosen namely (non-informative prior density function) as a result of
this study we obtain the following result the estimation regression model is equal in all
methods and Baysian estimation method is the more efficient way in estimation and
(Ordinary least squares method) with (maximum likelihood method) are equal in
efficiency.