2011 / 

         7 

 4 

  
  
  
     
 
 !
Using Seasonal Time Series Models to Forecast Electrical Power
Consumption in Fallujah City
*) & ) .. $% & '& (
 .."
( !) 
-+ .+ / 
+ )

1
  
   
  

 
 
    
  ! 
 $   
 - .   
(2010+2005) '(
)(
  ! " #
 $ %
 &
. 
 
  ! 


 
  
 
    0
1(#
 "
 
  2%
 3"  45
SARIMA (1 , 1 , 1) × (0 , 1 , 1)12 : )
  6 7

 #  >
 2011  0
  #  '(
9
 &:  # ;
  
  < 3" 
=<!
. @? 
 
 ! 0 A =<  <
 - 45  B 2012 ?

Abstract
This research deal with using seasonal time series models to study and analysis
the monthly data on consumption of electricity in Fallujha city for the period (2005-
2010) , whereas this models are distinct with high accuracy and flexible in analysis time
series .
The results of application show that the proper and efficiency model for
representing time series data are the multiplicative seasonal model of order :
SARIMA(1 , 1 , 1)×(0 , 1 , 1)12
According to estimation results of this model done forecasting to monthly
consumption of electrics capacity for two years ahead from the period Jan. 2011 to Dec.
2012 , these values show a harmonic direction with the same original time series.

2
2%   E  ):  @$:  
 
 !  C
 - ): D@5 <

 @$: 4

 C
   ): F G" @
 
&
 " @E H $E "  #
 I <

.  J 7 I 5 K 4
  
 
 # 
  $C
  E  ):
I <
  I 5 K 
  
  <

 L
? 2%
  C
 %C 1 @E 
L
? 5   
 
 C . 
 H7 4
'
 >
  C
 #

21
  2011 /           7 
 4 

 
 A ' 4
 $ K 5% 
 >
 C% !C 
  # 
 0
 " @E
<
 ! ' 4
 $ K >    ;
  ' 
  $C
 ( 3" 
 L ?
. 7 
 ! 
  17 K
#)+I#   MC ) 
 
 
    ! 
  5 1 ) &


<$  ; %%
 N
! B &: <   B " #
 $ %
 &  ;
( 

. H %<
  !  
 <
 %%
 A7 K = $ .C)
7: 3% 45
$ %
 &  ;
. 
 
 
 
O(#: 7!? 
 
. 2012 ?  #  >
 2011  0
  #  '(
)(
  ! " #

:  (  5 7!  
 2% 7: 3% 8
<   (2010+2005) '(
)(
  ! 9
 " #
 $ %
 &  +1
. = ' "    #
$ %
<
 ) : = 5 
=  )(
 
" #
 $ %
 &  ;
 C +2
.  : 4!  N K #

7: ) 9: 3%  
L )
 ! 

  3 B 94
 L )
 ! 
 (@
 3
  3 
 
K
 % #  2%
 .!  9
 94
 L )
   ) K
 
 <  <! &

 . <%

2% 
 1    C
 #
   
 
 
   @ 
 4
 I?
 C$    K ( 
  ) <%  1) .!  <! <%
 L )
 5 . ;
 <

" #
 $ %
 &  ;
7   K
 @
)(
  ! " #
 $ %
 &
? 5 B  @
 2
  @
  ) : 5 K ? 1)
 7 B.<: (

. Minitab 
 3  SPSS Ver.17 " @E 3 
 ! 

7: $( ;
 71
;
 
  
 $C
  0 
 5 <! &

 >
 5 2$
 ;

05 
 (Box & Jenkins (B-J))   5   # ;
  
 2%
  C
 C7

 
      
  1)
  ! &

B <% : ) ! $ " (#
: (B-J) L5 C "   

Seasonal Time Series :    7171

  A L$ C #  
 7C A %
 ' 
 <
  )  @<
  <
  ? ! 
 
 #
0 
 %
 K
   
 
 ' 4 K 9
 0  ! C ( C 
  95 B (Brock Well & Davis , 1991 :53) 

 


22
  2011 /           7 
 4 

5 
= @! 5  (s) # $ (s) 
 
 
 '(
 '(
 - K (Fixed intervals)
 95 B =
f(t + s) = f(t)
 2%   ( # # 
 -  C
 - ): A )  #  
  LC@


 '< J  C   
 !  C
 - ): )  95 B '<   
 #  

. 2(
   '<    K
  #
(

  % : G! 2%  
   '<    K @  C
 C 2! 
 % : &  )  ! B (Makridakis & McGee , 1983 : 263) 

Anderson , )  # '<
 
 
  ! ( 
 % 0 )  0  !
 
 
  6@ !  
 " @E  C
 MC ! . (1976 :31
:   0 B )

Autocorrelation (AC) :  

+ : 271
 F !  .( >
 $  $C
 )
I < .  
 % :  C 6C
>@
 L (S) E  
 
 
 
 ! 
 % :  C < B (
: (Wei , 1990 : 21) :
n -S

Cov(Z t , Z t +S ) ∑ (Z t - Z)(Z t +S - Z)
ρˆ S = = t =1
n
.............................(1 - 1)
Var(Z t ) Var(Z t +S )
∑ (Z
t =1
t - Z) 2

. 
 
   $ : Zt :  


 % 
:  
C  4
 

 

 

 

 
!
 E  %<! .5 B (@
   $ $ # ρ1 , ρ 2 , ρ 3 , ....... , ρ S −1 , ρ S , ρ S +1 , ......
. 
 % :  C
C $ ) k=0 , s , 2s , …..
  C %C : 
 
 
  !  (ACF) 
 % : 
 
# . : 5    
 <  K   # B  ?  #  
 &
" G(
Residual Autocorrelation Function (RACF) $

 % : 
 

(ACF) 
 % : 
  !   (@ . ;
 1 %5 "    2%  

. 1 %:  )  # (@
  $ #  (G$ )   '<


Partial Autocorrelation (PAC) 7:     

+ 7: 371
< 0 Zt+S Zt  
  $C
 )
I < .  ")
 
 % :  C 6C

  5 < : (PACF) ")
 
 % : 
  . Zt+1 , ……. , Zt+S-1 N?  


23
  2011 /           7 
 4 


 G  ! =7  
 
  !  '5 =7 ! (ACF) 
 % :
Wei , 1990 : ) ( $
) ;
 1 % "      
 " G! ! .) 
. (23
Seasonal Time Series Models :    
 7: 471
:  %+  :17471
Seasonal Autoregressive Model (SAR)
  )
) ? # 
 O (p) )
  
 
 : 
7 
 >@

:(309+281: 1985 B  C

Z t = Φ S Z t -S + Φ 2S Z t -2S + Λ Λ + Φ PS Z t - PS + a t ......................(1 - 2)
: 5 
i=0 , 1 , 2, ……, p B 
 
 
   $ : Z t-iS
. 
 '(
 % : S
i= 1 , 2, ……, p B 
 
 : 
C : Φ iS
. 
 
 ) : p
a t ~ NID ( 0 , σ a2 )   B " C
 O%
 : a t

 C
 ) #  %  <: 2< #

Φ S (Β ) = 1 - Φ S B = 0
S S

=< 
 # #
. 95 B (  9  %$ 6@ '" ) unit circle '
 '"  
: (Wei , 1990 : 161) #  % 
− 1 < ΦS < 1
: #
 6C Back shift operator (
 :    B  
Β S Z t = Z t -S ∀s = 1 , 2 , Κ Κ
SAR(1) K
? )
  
 
 : 
(ACF) 
 % : 

 C
 >@
 
: : #
 O
1 k=0

ρ K = Φ S k =s
0 k = 1 , 2 ,Κ Κ , s - 1

")
 
 % : 
   ! B =5 G$  AR(p) 
 

 % : 
  95
. ( Makridakis & McGee , 1983 :264 ) p @ (
 '(
 C A%<
7:  % 
:  7: 27471
Seasonal Moving Average Model (SMA)
: : >@
 ! (B) (
 ( : ) E    

24
  2011 /           7 
 4 

Z t = Θ (Β S ) a t
S

= (1 - Θ B S - Θ B 2S - Λ Λ - Θ Β QS ) a
S 2S QS t
  )
) : #
 O (Q) )
  
 #
 % ? 
 C
 >@
  !
: (309+281 : 1985 B  C

Z t = a t - Θ S a t -S - Θ 2S a t -2S - Λ Λ - Θ QS a t -QS .....................(1 - 3)
:  
. 
 #
 % ?  
C : Θ iS
-1< Θ < 1  i = 1 , 2 , Λ Λ , Q
. 
 
 ) : Q
 ! . ( Q )
 C (@
 K
 ; ) Qs '(
 C A%< (SMA) 

 % : 
 
. =5 G$  (PACF) ")
 
 % : 
 
:  ( % 
: –  %+ )   7: 37471
Seasonal Mixed (Autoregressive – Moving Average) Model (SARMA)
: : >@
 ! (B) (
 :    
Φ S (Β S ) Z t = Θ S (Β S ) a t
( 1 - Φ S Β S - Φ 2 S Β 2 S - Λ Λ - Φ P S Β P S ) Z t = ( 1 - Θ S Β S - Θ 2S Β 2S - Λ Λ - Θ Q S Β Q S ) a t

:(13 :1996B97
) ? #
 O (P,Q))
  
 %
 
 C
 >@
  !
Zt = Φ S Z t -S + Φ 2 S Z t - 2 S + Λ Λ + Φ p S Z t -p S + a t - Θ S a t -S
- Θ 2 S a t -2 S - Λ Λ - Θ QS a t -QS ...............(1 - 4)
ARMA(P , Q)S 
 .
 9

. ! '< J (Zt) 
  #   B '<
 
 
 K 2% -5 
  
 

 2(
     . '<   : %
 
 2(
  ) C 
  ) #
 D )
 
∇ =(1- Β )
D
S
S D

<
 J 
 %
 
 
# #
) : #
 L# 9
 ( PSeasonal Autoregressive Integrated Moving Average Model )
: ( Box and Jenkins , 1976 :154
Φ P (Β ) ∇ Z t = Θ Q (Β ) a t
S D
S
S
...............................(1 - 5)
ARIMA(P , D , Q)S = @ L# 9
 
 )  ( P , D , Q )  
7: 4&
8   7: 47471
Multiplicative Seasonal Model (SARIMA)

25
  2011 /           7 
 4 

B 97
 )  ( p , d , q )*( P , D , Q )S )
  6 7
 
 
 C
 >@

: ( Anderson , 1976 : 54 ; 15 : 1996
φ p (Β ) Φ P (Β S ) ∇ d ∇ SD Z t = θ q (Β ) Θ Q (Β S ) a t ...............................(1 - 6)
:  
. 
 J 
 :  ) : p
. 
 J 2(
 ) : d
. 
 J #
 % ?  ) : q
. 
 J 
 :  C : φ p (Β )

 ∇ = 1 - Β   d 
  
 J 2(
  C : ∇ d
. <: 
 K
 <:  
  
 

. 
 J #
 % ?  C : θ q (Β )
. 
 
 :  ) : P
. 
 2(
 ) : D
. 
 #
 % ?  ) : Q
. 
 
 :  C : Φ P (Β S )

 ∇ S = 1 - Β S   D 
  
 2(
  C : ∇ SD
. <: 
 K
 <:  
  
 
 

. 
 #
 % ?  C : Θ Q (Β S )
(0 , 1 , 1)*(0 , )
  6 7
 
 
 C
  <%
 ! : C"
  
 
: ( Box and Jenkins , 1976 :173 ) : #
 O 
 
 C
 >@
 1 , 1)12

∇∇ 12 Z t = (1 - θ 1 Β )(1 - Θ 12 Β 12 ) a t .......................................(1 - 7)
(1 - Β )(1 - Β ) Z t = (1 - Θ 12 Β - θ 1 Β + θ 1 Θ 12 Β 13 ) a t
12 12

Z t = Z t -1 + Z t -12 - Z t -13 + a t - θ 1a t -1 - Θ 12 a t -12 + θ 1Θ 12 a t -13

- 1 < θ 1 , Θ 12 < + 1  
2(
 5 C (@
 J $ 
  % : 
 
 
 

  
2(
 5 C ) 
 2(
   K
 ' E ) . (1 , 11 , 12 , 13) 
 (
 ∇∇ 12
 %
  
   (C)   (yt : t=1 , 2 , …., N-C) #
  C ∇∇ 12
.(13) 

 - ! 9  

Testing Stationarity of Time Series :   2 
 : 571
5 '< 
 - 5       
  @$: <%
  
 4C M(
 ) <: C 6@  @$: ' 
 @ 
 
 
 LJ5   ! B # 

26
  2011 /           7 
 4 


 (ACF) 
 % : 
 
 
      # B %
 64
 <
<:  A) . 
 <   < K #
 (PACF) ")
 
 % :
:( 614 : 2000 B %) 

 L ? ?
 
 %
  
 <  * .  < ) * .   - ) ) *
.  0 
 
 
  
 %
 # O " @E  
  <:  @<
: (25+3 : 2003 B  >
 )   
 
 < 0 ! 
 L
?  
7: =
 9
> & 
%  71
B 
   
C%
  J
 K @
 B 
   0 ! 
 
 5 
.   
 L< 5 
C
 )
 K @
 5
7: 
) ?
+ 
%  72
:   #  C
 - ):  G

 2%
 
'<  # $
 A  C
 .
 0  C
 - ): < ! %
 : <% +5
. detrending C
 - K
B C 1 % (
MC
 7C   
 $ R% <%
 - 7< : 2(
 <% +L
: #
 O K
? )
  2(
 
= 0!
y t = ∇Z t = Z t - Z t -1
: : #
 O  0
 )
  2(

y t = ∇ 2 Z t = ∇Z t - ∇Z t -1
= Z t - 2Z t -1 + Z t -2 = (1 - Β ) 2 Z t
. '<   K @
 C
 - ):  G
2(
  (d) 2% K
  
 O) $
7: (  = 1 )  9
2 @ 73
Seasonal differencing 
 2(
 <%  
 @C
  
 
 )

: 2000 B %)
= 0!   
 H A <
 1 %E ! L MC
 7C  <
 R% &

: (631
y t = Z t - Z t -4  A 2(

y t = Z t - Z t -12 
 2(


E B Zt K @! 
C
 )
 5  
 0
 yt     
 M! K
K @ 
  <
 
E . Ft = Z t - Z t -1   Ft 
 K @  C
 - ): 05
. Wt = Ft - Ft -12 K @! Ft 
=   0 '
K
? 2(

Stages of Building Seasonal Model 7:   A
 % 7: 671

27
  2011 /           7 
 4 

)  B  (@  =$% 
 
 4  K C Box & Jenkins )
)   CO  B 
 ' 4
"
 
  ARIMA   "  7 
: ( Box and Jenkins , 1976 :243

Identification : 1B 71

 0
L 
 
   5 
 
 
 ! <: 2< C
K <%
 - C . (PACF) ")
 
 % : (ACF) 
 % : 
   .)

")
 
 % :  C < %   (PACF)  (ACF) 
 
 
 $
: )
 ! D7
 ")
 
 % : 
 % : 

94
 &
 A 
 

. (18: 1996 B 97
 )
.  
+ C%
D2  ) : (1)   

 (ACF) 
 % : 
 ")
 
 % : 

(PACF)
&  =5 =# .#
 =) G$  PS 
 E C A%$
SAR(PS)
( =) K ) L)
 
 (Cuts - off)
(Decays Exponentially)
QS 
 E C A%$  =5 =# .#
 =) G$ 
SMA(QS)
(Cuts - off) ( =) K ) L)
 
 &
(Decays Exponentially)
&  =5 =# .#
 =) G$   =5 =# .#
 =) G$ 
SARMA (PS , QS)
( =) K ) L)
 
 ( =) K ) L)
 
 &
(Decays Exponentially) (Decays Exponentially)
Estimation : 2 72
6 
 <
   
 <
 2"% N   .
C <  "
 
  C
: ( Lawrence & Paul , 1978 : 629-642)  
 
 L
. (%7
)  
 4?  #E <% +5
Exact Maximum Likelihood Method (EML)
. %
 J N>@
  C
 <% +L
Non Linear Least Square Method (NLS)
Diagnostic Checking of Model :    1% 73

 
 
    0

 @ 5 " N    : 
 < C
: ( Wegman , 2000 : 443 )  2% ' &

)

28
  2011 /           7 
 4 

&

 B =C (@
  6 95 " @F C  #  : 
  C +5
. MA 5 AR L 5  C  : C J  #  ! (t)   
: :  : &

 Residual analysis $
  +L
Confidence Interval Checking +: <0
 9   +1
B: 5 (@
  =C 6 ( rS(a)) s 
 E  1 %S

 % : #  :
  . (0.95)    ( µ 1.96/ n ) <0
 9  A<  L) .$  !
rS (a) - o
Zt =
1
n
 !
  1   1  
Pr - 1.96   ≤ rS (a) ≤ + 1.96    = 0.95 ............(1 - 8)
  n   n  
# (1(#) " ) 
  ="  H ( $
 ) 1 %?  C ! &
 2< 
(@ -<   % =C% H < # $

  % :  ;
 ! .
 95  1 
  -$  
 n 
1
rS (a) ~ NID ( 0 , )
n
+: Portmanteau   +2

 ( Pierce & Box " @F ) Q 1 @E  
 " G(
= 0#?  : 
Box & Price , 1970 : ) : >@
 2! $

  % 
" @E C
  : 
: ( 1509-1525
L
Q = n ∑ rk2 (a) ~ χ ((L
2
- m ) ,α ) ............................(1 - 9)
k =1

 
. '<
 
C
  : m T 
  E  : L
J 
  % :  3 Ho C
 7! < 
)
 χ 2 $  >@5 Q $  #  !
<
 . " ) 
 2!  #;  < #  H "  $
  K
    C
: 1: >@
 O
LJung and Box $  >@
 - % C 
L
rk2 (a)
Q* = n(n + 2) ∑ ....................................(1 - 10)
k =1 n-k
χ ((L
2
- m ) ,α ) A =7 A " @E -

. GH‫ا‬JKLM GNOPM‫ ا‬GQ‫ا‬RM‫ط ا‬TKQ‫ر‬V‫ ا‬WLX YZ[\Q ‫ر‬TK[]^M ‫ أ]`ى‬bcd YeJQ *

29
  2011 /           7 
 4 

-   7!? 

    : B =" @   ' $ 
 ! .  #
 )
 
: (Akaike , 1974 : 716-723) : 7 (
  C
=<!  

. (C7 $  
   #  +5
.
= "7 $
  C H) #  +L
" N5 ' C 5 B 
= "7  
<<
 ! 0#
  
 ! 0#  2 (
 #  +
=7  6C  Akaike   C
  B '<
 
C
  '   < 
  
: >@

2(p + q)
AIC = Log (σ 2 ) + ...................................(1 - 11)
n
. 
   : σ 2 :  
. '<
 
 
C  : (p+q)
: 1998 B    ) >@
 2!   
   #? C
  
#  ." % L
: ( 173
AIC
MAIC = .............................................(1 - 12)
n
: : C
 Schwartz 6 75
p+q
BIC = Log(σ 2 ) +   Log(n) ....................................(1 - 13)
 n 

Forecasting :  74
! .  
 " G! <
 G 
    "
 
  C
@
(t) 
  %
 A$
 O &
  $ '! (L= 1 , 2 , …..) K
 <
 <
 ;

. #  $5 ;
 O%  C % Zˆ (L) = Z ;
 K
t t +L

Zt 
<  .
 $ K N 
 Differences Equation Form 2(
 
 C >@  
) : >@
 2! 
 %
 
;
 L  # (at) " C
 O%
<  
 $
. ( Box and Jenkins , 1976 :289
ˆ (L) = Φ
Z t +L = Z t
ˆ Z ˆ ˆ ˆ
S t + L - S + Φ 2 S Z t + L - 2 S + Λ Λ + Φ p S Z t + L -p S + a t + L - Θ S a t + L - S -

ˆ 2 S a t + L- 2 S - Λ Λ - Θ
Θ ˆ QS a t +L-QS ............................................(1 - 14)

:  
a t +L = E(a t +L ) ; Z t + L = E(Z t + L )

30
  2011 /           7 
 4 

7: 9

 4- 7: 172 : 2 ;
 72
&: 0 '  (72) A$    6
; 
  !  
   
 
) &: 6 @5 A)
  / % #
 '<
 )(
  ! " #
 $ %
C(
 9

A '" )  5 
 (...... B 1 (E H
 ' F B #
 B 9 )
 B


?  # K
 (2005)  0
  #  '(
 
 B (2) $ )
 ! # B )(
 1 #
(166.300) N@$ $ (2005)  ! ) (95.000)  $ (133.414) -$ % B(2010)
%  
 - $   . (%0.76) -$ 9  C  (2010)  )
 . 
 
    I )  )  '#! %C   (23.714) -$ 9 C 6 
.) 5 K 
 G  #

 =C% =C A 
  M!: ! #  
 
. (201072005) .!  !   
  
 = $ B  + : (2)   
year 2005 2006 2007 2008 2009 2010
Month
Jan. 97.950 99.910 144.000 160.000 160.500 161.785
Feb. 97.330 98.650 139.500 146.000 147.000 148.730
Mar. 95.000 122.300 138.000 142.000 144.000 142.610
Apr. 95.200 122.000 138.000 145.000 146.000 159.250
May 95.000 122.000 138.500 146.500 147.750 161.107
Jun. 96.000 124.000 139.000 146.000 147.000 165.752
Jul. 97.000 130.520 140.000 150.000 152.000 162.230
Aug. 97.300 130.500 144.500 155.000 156.000 165.540
Sep. 96.270 96.000 144.000 155.500 156.500 157.600
Oct. 96.000 96.000 145.000 158.000 159.000 142.800
Nov. 96.000 96.100 128.750 130.000 130.500 127.500
Dec. 97.000 98.450 140.600 145.000 147.000 166.300
. )(
 1 # A '" : @

7:   % 7: 272
7:    7: 17272
D7  # B(2) $ )
 ! 
 
      
 
  1
 $
A    - ) ) #
  4 
? @" @ K 6C
. 2
 ! (1) $ #
 !
# '
 I(  4  #  
 - B; C< ! 0   )  
= 7! 

- ) # ) K 
 ; >
 - B N5 K
     
 '
 6 A 
.  #  
7:   2 
 : 27272
O ' (Ln) C%
  J
 O   
 )
  
 ! <: K @
 M>

 )  # : - 4  . &

  2
 ! (3 B 2 ) $  # : N5 ' C
 )


31
  2011 /           7 
 4 

2%    :  &


  
C%
  J
 O   K
 << $  
 ! <:
.  

!C 6 &
 #O
   
 ! =   ) &   ) ( 2  1 ) $ #
  
 ! 4 
 (4) $ #
 ! # ")
 
 % :  C   
 C%
: 
 % :  C  B (@
  =C 6 (18) ')(
 K 
 % : 
  C
#
 C
  : LJung & Box     ( - 0.23 ≤ rk ≤ + 0.23 ) <0
  7 
*
2
Q .stat = LB Q = 387.11 > χ (18 , 0.05) = 28.87  ) 
 % : 
  C

=(@ 9    

 % : 
  C #   " <
 C
 7! M! &


H0 : ρ1 = ρ 2 = ρ 3 = Λ Λ = ρk = 0
. '< J 
 
  C  
 7(
 < .
7:  2 & @ : 37272
7: 
) ?
+ @ 7"
 
C
 
 K @ K
? )
  2(
 5   C
 - ): 
F )5 
5 C 
C
 
 
 K  (5) $ #
 B (  J
 
) ∇Z t = Z t - Z t -1 
 C
 - ): L J K   @(
  9 
  ) #
 4  . 
? 2(

LJung & Box " @F 
-#;   '< J 
  95 
 #
 1 < A 
 !
 
*
Q .stat = LB Q = 28.97 > χ (17
2
, 0.05) = 27.59

J 
 
  ! . 
 % :  C # C M( 
 C
 7! M! &


. W7 '<
7:   @ 7;
! 7
 X 
? 2(
  C 
C
 
 

  % : $ 4 
95 B  
 
  K   (24 B 12) (
 ! C <
 -   (6) $ #

)
  ( 
 ) 2(
 5  
  G
 M>
 &

.  (12) # ( C 
  (C4) 
C
 
 K @!   0
C4 = ∇∇ 12 Z t = Z t -1 - Z t -12
 )  B ( ∇∇ 12 Z t ) 
 2(
  C 
C
 
 
 K  (7) $  
 #

B

 K ( 9 B 8 ) $ #
 ! #   ")
 
 % : 
 % :  C $
C ( - 0.23 ≤ rk ≤ + 0.23 ) <0
  7  
 % :  C  (8) $ #
  4
. 
 
 < K   B '   0
 '(
 ! %<! C  (12) 
 E

32
  2011 /           7 
 4 

7: 1B : 372

 # K  :  &
 MA AR  
     
 K 6C
 C
")
 
 % : 
 % :  C $ < %  B (Conelogramme) 
 % :

94
 &
 A (9  8)  # : ! # 

 K
? 2(
 5 C 
 

C
(PACF) ")
 
 % : (ACF) 
 % : 
  D7 (1) $ )
 ! D7

  3  ;
    ( L)
 
 & & ) (k) E ! '  A =) G$ 
)
  6 7
 
 
  

SARIMA ( 1 , 1 , 1)×( 0 , 1 , 1)12
or
(1 - φ1 Β )(1 - Β )(1 - Β 12 ) Z t = (1 - θ 1 Β )(1 - Θ 1 Β 12 ) a t

: 2 : 472
SARIMA ( 1 , 1 , 1)×( 0 ,

 "
 
 K
 @ #
  
  C C
<% 2% .  
 I )   
C
 C AKIAKE  C K  :  &
 1 , 1)12
  
 $  J
 
 
    K (NLS) %
 J N>@
  C

: : 3" 
 K @
  SPSS Ver.17 Minitab " @E 3 

Final Estimates of Parameters

Type Coeff. StDev. T

AR1 0.8140 0.0860 9.47
MA1 0.9718 0.0291 33.40
SMA12 0.7027 0.1608 4.37
Differencing : 1 regular , 1 seasonal of order 12
Number of observation : original series 72 , after differencing 59
Analysis of Variance :
DF Adj.sum of squares Residuals Variance
Residuals 56 0.334307 0.0059697
Standard error = 0.077264
Log Likelihood = 61.706898
AIC = -2.122352
MAIC = -0.035972
BIC = -2.134004
. ((@
  =C 6 ) " @E  
  ) 
C
  -5 3" 
  4
:    1% : 572
&
  " (# 
 " @  #O
  : -< .)  
 G  C
:    
: $

 % :  C   +5

33
  2011 /           7 
 4 

! #   <

 
(1 %?) $
")
 
 % :  C  
<0
  7 A< $

 % :  C $ A)  . 4 (10) $ #

. " ) 
 
  "  $
   C  ( - 0.2552 ≤ rk (aˆ ) ≤ + 0.2552 )
Portmanteau :   +L

 % :  C  a t ~ NID ( 0 , σ a2 )  95 (White Noise)  1 %?  
1
(LJung '1 @ 2%  <! ( ) -<   (@ 9    % =C% H rk (aˆ ) $

N
  4 
 " G(
& Box )Q*
*
Q .stat = LB Q = 10.6 < χ (29 , 0.05) = 16.92
. 1(# " ) 
 2!  ! 0  (" ) C J $
   &
  3
: <
 
 $
C%
 A
   +
 #O
  : 
C
 K  (t)    !  #O
 ;
 <0
  L 

C%
 D%
 4 
  7!     (11) $ #
 ! D7
 1 %S
C%
 A

. C%
 A
 G" @  $
   K  $
 

. ;
 ! . #

 =" @ 
 $ K
 9;  : - #
7:  : 672
 9
 &:  # ;
  -5 (4+2) '<(
 ! . @
 ;
   
<
  C (3) $ )
 ! 3" 
 7 2012 2011 
)(
  ! " #
 $ %

4  B (12) $ #
 ! # ;
 -

 
   B @? <
 K
  J

. @? 
&
 I( A  O
 '(

  =) 

(3)   
. 2012 2011   !  
  
 =
 E $ B  + 9

year 2011 2012
Month
Jan. 186.757 190.844
Feb. 172.604 177.734
Mar. 170.088 176.241
Apr. 175.398 182.664
May 175.568 183.598
Jun. 176.191 184.869
Jul. 177.506 186.759
Aug. 180.534 190.366
Sep. 171.935 181.629
Oct. 167.487 177.191
Nov. 147.104 155.815
Dec. 168.351 178.493

34
  2011 /           7 
 4 

7: 9
- 9

+ 73
: 

 3" 
 G # <  : 9

+ : 173
3
 %%
 .) ! - ! " #
 $ %
  9
 &:  ;
 5 # +1
! '    K
 %%
  K
 9; )
 ;
   B ;
    

.  :  # 2C
7
 >
  ! #
  C
 !   >
  
  $C
 L J  +2
.;
  ! 2? C 
 
 L5  ! B
D7   - ) &    
 ! '< J 
 
  " @E  :  +3
 B = (12) # ( C   
 #
 K "  
= 7! 
 !
 C
 - ): 
F  
 0 =:5 C $ 
 ! <: % ! )5
5 C 
 #
 
Y =  0   
   J
K
? )
  2(
  
.(12) )
  2(

B 
  
$ $5 ) 7 (
  C   #
  
    7!5    +4
=" @ R<
 
 " G!  B ( BIC B AIC B $
  C H)
$ $5
C%
 A
 B $

 % : 
  B '<
 
C
 C :    
. $

6 7
 
 
  
 
    0
1(#
 "
 
  ) +5
. SARIMA ( 1 , 1 , 1)×( 0 , 1 , 1)12
(24) '(
)(
 
" #
 $ %
  9
 &:  # ;
  
 
=<! +6
$ B @? 
 ! 0 A =<  <
 - 45  . 2012 2011 
=
. 
 ! 1 #
 & A$
< '@ 

7: 9
- : 273
:   @ 
 @
  
 3" 
  
C
 L? - : $C
   )
 $  ;
'C
 >@
 
  3"  ? +1
. ;
 ! "

N K N? 7$: 4! 
 N K '4    K
 
  C +2
.   < 1) N?  4! 


35
  2011 /           7 
 4 


-
– I#   5 2% " B (1985) B K7 M  B  C
   )
  0 B  )
 +1
)
 B !
  ) B " @
  ! '
  ) ;

 
#)
. (309+281)G B   I 
 C
 B A 

  MC
;
 <
 2"%
 <  " B (1996) B  9$  B 97
 +2
. > C ) –  @$: 'E # – 1 @F )  
 B " 
 #) – I#
C
 #
 ! ?  C5  ;

 
  " B (2003) B \   B  >
 +3
B  0
 C
 B C
 &
 C ) ) B " Box – Jenkins )   C

. (25+3)G
.C )
  %
  B ")
 B " N
 @$ ;
   " B (1998) B 
 B    +4
 B #: B "  @$:  $C
 I $ 2% " B (2000) B   <
 B % +5
. @
  C )

.C )
  %
  B ")
 B "  
 2! 1 @E " B (2000) B  B 
+6
7- Akaike , H. (1974) , "A new look at the statistical model Identification " , IEEE
Transactions on Automatic Control , Vol.19 , No.6 , PP. 716-723 .
8- Anderson , O.D. (1976) , " Time series analysis and forecasting " , Butter worths
, London and Boston .
9- Box G. , E. , P. , and Jenkins G. , M. , T. (1976) , " Time Series Analysis
Forecasting and Control " , San Francisco , Holden-Day , U.S.A.
10- Box , G. E. and Price , D. A. (1970) , " Distribution of Residual
Autocorrelations in Autoregressive – Integrated Moving Average Time Series
Models " , JASA , Vol.55 , No.332 , PP.1509-1525 .
11- Brock Well , P.J. and Davis , R.A. (1991) , " Time Series Theory and Methods "
, 2nd ed , Springer Verlag New York Inc , New York .
12- Lawrence , A.K. and Paul , I.N. (1978) , " on Conditional Least Square
Estimation for Stochastic Processes " , Ann. of Stat. , Vol.6 , No.3 , PP.629-642
.
13- Makridakis , S. , Wheel Wright S. , C. , and McGee (1983) , " Forecasting
Method and Application " , 2nd ed , John Wily and Sons. Inc. , U.S.A. .
14- Wegman , E.J. (2000) , " Time Series Analysis – Theory , Data Analysis and
Computation " , Addison-Wesley Publishing Company .
15- Wei , W.S. (1990) , " Time Series Analysis : Univariate and Multivariate
Methods " , Addison – Wesley Publishing Company Inc. , U.S.A .

36
 2011 /           7 
 4 

F%

2010+2005 '(
" #
 $ %
  9
 &: K : (1) $ #


. 
C%
  J
 5 C 
 
    0 : (2) $ #


37
 2011 /           7 
 4 


C
 )
 5 C 
 
    0 : (3) $ #


Autocorrelation Function for C2

1.0
Autocorrelation

0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

2 7 12 17
Lag

Lag Corr T LBQ Lag Corr T LBQ Lag Corr T LBQ

1 0.86 7.28 55.27 8 0.53 1.65 276.54 15 0.26 0.69 372.87

2 0.77 4.14 100.13 9 0.50 1.48 297.29 16 0.24 0.63 378.21
3 0.72 3.18 139.62 10 0.46 1.33 315.49 17 0.22 0.59 383.07
4 0.65 2.55 172.60 11 0.45 1.28 333.24 18 0.20 0.53 387.11
5 0.60 2.18 201.68 12 0.43 1.19 349.70
6 0.57 1.93 227.81 13 0.33 0.89 359.41
7 0.55 1.79 252.85 14 0.28 0.76 366.68

38
 2011 /           7 
 4 

Partial Autocorrelation Function for C2

Partial Autocorrelation 1.0

0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

2 7 12 17 Lag

Lag PAC T Lag PAC T Lag PAC T

1 0.86 7.28 8 0.02 0.19 15 0.04 0.31

2 0.12 1.00 9 -0.05 -0.39 16 0.08 0.67
3 0.13 1.07 10 -0.02 -0.19 17 -0.00 -0.01
4 -0.03 -0.25 11 0.08 0.72 18 -0.04 -0.34
5 0.06 0.53 12 -0.00 -0.00
6 0.03 0.26 13 -0.32 -2.71
7 0.10 0.82 14 0.04 0.33

. 
C%
  J
 5 C 
 
")
 
 % :  C : (4) $ #


. 
: 2(
 5 C  J
 
 
 K : (5) $ #


39
 2011 /           7 
 4 

Autocorrelation Function for C3

1.0
0.8
Autocorrelation 0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

2 7 12 17 Lag

Lag Corr T LBQ Lag Corr T LBQ Lag Corr T LBQ

1 -0.13 -1.06 1.16 8 0.07 0.58 7.68 15 -0.02 -0.11 28.60

2 -0.23 -1.90 5.11 9 0.02 0.19 7.73 16 -0.03 -0.23 28.72
3 0.02 0.12 5.13 10 -0.14 -1.06 9.33 17 0.05 0.33 28.97
4 -0.13 -1.00 6.37 11 0.01 0.09 9.35
5 -0.03 -0.23 6.44 12 0.41 3.10 24.10
6 -0.10 -0.76 7.19 13 -0.21 -1.41 28.06
7 0.02 0.15 7.22 14 -0.08 -0.50 28.58

. 
: 2(
 5 C 
C
 
 

 % :  C : (6) $ #


. 

 K
: 2(
 5 C 
C
 
 
 K : (7) $ #


40
 2011 /           7 
 4 

Autocorrelation Function for C4

1.0
0.8
Autocorrelation

0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

4 9 14 Lag

Lag Corr T LBQ Lag Corr T LBQ

1 0.01 0.05 0.00 8 0.24 1.60 14.18

2 -0.22 -1.67 3.00 9 -0.01 -0.05 14.18
3 -0.00 -0.03 3.00 10 0.23 1.50 18.17
4 -0.28 -2.06 8.16 11 0.03 0.18 18.24
5 -0.08 -0.52 8.54 12 -0.37 -2.27 28.52
6 -0.15 -1.00 9.99 13 -0.06 -0.33 28.78
7 0.05 0.35 10.18 14 0.04 0.21 28.89

. 

 K
: 2(
 5 C 
C
 
 

 % :  C : (8) $ #

Partial Autocorrelation Function for C4
Partial Autocorrelation

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

4 9 14 Lag

Lag PAC T Lag PAC T

1 0.01 0.05 8 -0.04 -0.33

2 -0.22 -1.67 9 -0.07 -0.52
3 -0.00 -0.01 10 0.20 1.51
4 -0.34 -2.65 11 0.02 0.12
5 -0.09 -0.66 12 -0.23 -1.74
6 -0.37 -2.81 13 -0.05 -0.38
7 -0.02 -0.17 14 0.06 0.47


 K
: 2(
 5 C 
C
 
 
")
 
 % :  C : (9) $ #

. 

41
 2011 /           7 
 4 

ACF of Residuals for C2

(with 95% confidence limits for the autocorrelations)

1.0

0.8

0.6

0.4
Autocorrelation

0.2

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Lag

PACF of Residuals for C2

(with 95% confidence limits for the partial autocorrelations)

1.0

0.8

0.6
Partial Autocorrelation

0.4

0.2

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Lag

. <
 
 $
")
 
 % :  C : (10) $ #


42
 2011 /           7 
 4 

Histogram of the Residuals

(response is C2)

20
Frequency

10

-0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Residual

<
 
 $
C%
 A
 : (11) $ #



 
<
 ;
 <
 
 K
 : (12) $ #


43