Statistical Forecasting Methods

) (

.
.

.

.

ARIMA
Time
Series

Minitab

.

Intervention Analysis
Transfer Function Models
Multivariate Time Series Models
Vector Time Series Models State
Space Models and Kalman Filtering Threshold Time Series
Models ARCH GARCH
Finantial Time Series Forecasting
Neural Networks .

.

http://www.abarry.net/or/or221book1.pdf
.
.

1422
2002
-1 :
10..................................................................
1.1
10.......................................................
1-2
10 .. ......................................
1-3
10 . ... ................................................
1-3-1
10 ..... ..................................................................
1.3
11 ........................................................................
1.4
11..............................................................
1.5
11........................................................................
1.6
11 ...................................................
1-4
11.....................................................................
1-4-1
11 .............................................
1-4-2
11 .........................................................
1-4-3
11 .........................................................
1-4-4
12 .............................................................
1-4-5
12 .................................................................
1-4-6
12 .........................................................
1-4-7 : 1
12 ...................................................
1-4-8
13 ....................................................
1.4.9
13 ...................................................
3
1.4.10 : 2
13 ................................
1.4.11
14 ...........................................
1.4.12 :3
15............................
1.4.13
16..............................................
1.4.14
17 ....................................
1.4.15 :4
18 ...................
-2 : - ARMA
-
)p,q) ........ 22
2.1
22..........................................................
2.2
22 ........................................................
2.3
22..................................................................
2.4
22 .................................................................
2.5 ......................................................................
23 ...........
2.6 -
24 .......................................
2-7-1 ARMA(0,0) ...........................................................
24
2-7-2
AR(1) .................................................................................. 27
2-7-3 AR(2) ....................................................................
31
2-7-4 MA(1) .....................................................................
36
2-7-5 MA(2) ....................................................................
39
2-7-6 ARMA(1,1) .............................................................
40
2.7.7 ARMA(p,q) ................................................
47
4
-3 :
49 ...............................................
3.1
49 ..................................................................
3.2
50 ..................................................................
3.3 -- )
p,d,q) .................. 52
3.3 ARIMA(1,1,0) ....................................................... 52
ARIMA(0,1,1) ........................................................ 52
3.3.1

3.3.2
53 ......................................................
3.4 ) y ( B
ARMA(p,q) .................................... 53
3.5
54 .....................................................
AR(1) ....................................................
3.5.1
54
MA(1) ....................................................
3.5.2
55
AR(2) ....................................................
3.5.3
55
MA(2) ...................................................
3.5.4
56
ARMA(1,1) ...........................................
3.5.5
56
ARI(1) ...................................................
3.5.6
57

3.5.7
ARIMA(1,0,1) .................. 57
3.6 )y ( B
58 ....................................................
-4 :
ARMA(p,q) ....... 59
4-1 :2
61 .........................................................................
4-2 Information Sets .................................................
62
4-3 :3
62 ............................................
4-4
.......................................................................................... 2
62 ...
4-5
62 .................................................................................
4-6 ARIMA(p,d,q) .......................................................
62
4-6-1 AR(1) ............................................................... 63
4-6-2 63 ............................................................................
4-6-3 AR(2) .............................................................. 64
4-6-4 ARIMA(0,1,1) ................................................. 65
4-6-5 MA(1) .............................................................. 66
4-6-6 MA(2) .............................................................. 66
4-6-7 ARMA(1,1) ..................................................... 67
4-7 68 .........................................................................................
4.7.1 68 .....................................................
4.7.2 69 ........................................................................................
-5 : 71 ............................................
5-1 71 .......................................................................
5-1-1 71 ..............................................................................
5-1-2 d ................................................................ 71
5-1-3 p,q ............................................................................... 71
5-1-4 71 ...................................................................
5.2 72 ..............................................................................
5.2.1 72 ........................................................................
5.2.2 73 .......................................................
5-2-2-1 AR(1) .............................................................. 73
5.2.2.2 MA(1) ......................................................... 74
5.2.2.3 AR(2) .......................................................... 74
5.2.2.4 MA(2) ......................................................... 74
5.2.2.5 ARMA(1,1) ................................................ 75
75 ..............................................
5.2.3
76 .........................
5.2.4
5.2.4.1 AR(1) ......................................................... 76
5.2.4.2 MA(1) ....................................................... 77
78 .........................................................
5.3
5-3-1 78 ....................................................................
5-3-1-1 79 ..................................................
5-3-1-2 79 .................................................
5-3-1-3 79 ...................................
5-3-1-4 79 ....................................................
5-3-2 79 .............................
5.3.2.1 79 ...........................................
6
5.3.2.2 AIC ......................................... 80

80 ..........................................................
5.3.3
-6 : -- 113 ......
6.1 114 .....
6.1.1 SARMA(0,1)(1,1)12 ......................................... 114
115 ..................................
6.2
SARIMA(0,d,0)(0,D,1)s .............................................115 6.2.1
SARIMA(0,d,0)(1,D,1)s ........................................... 115 6.2.2
SARIMA(0,d,1)(0,D,1)s ............................................ 115 6.2.3
SARIMA(0,d,0)(1,D,1)s ........................................... 116 6.2.4
SARIMA(0,d,1)(1,D,0)s ........................................... 116 6.2.5
SARIMA(0,d,2)(0,D,1)s ............................................ 116 6.2.6
118 ....................
6.3
119 .......................................
6.4
123 ..........
6.5
6.5.1 SARIMA(0,0,0)(0,1,1)12 .................... 123
6.5.2 SARIMA(0,1,1)(0,1,1)12 .................... 124
125 ..........................
6.6
:
-7 : -
142 ............................................................................................
-8 : 155 ..................
-9 : 162 .....................
-10 178 ...........................................
10.1 181 .............................................................
-11 : 184 .............
-12 : 190 ...........
12-1 190 ........................................................................
12-2 190 ........................................................................
12-3 191 ..................................................................................
-13 :
198 .............................................................................
13.1 199 ...............................................................
13.2 201 ..............................................................
13.3 204 ...........................................................
13.4 210 .....................................................
219 ...........................................................................................

:1 Time Series
) (
:
-1 .
-2 .
-3 .
-4 .
:
-1 .
-2 .
-3 .

850
750
C
1
650
550
70
60
50
30
40
20
10
Index
:

.
:
: Model Identification
-1
Time Plot

.
: Model Fitting
-2

.
: Model Diagnostics
-3
Fitting Errors

.

.
: Forecast Generation
-4
Forecast
Errors
Control
Charts

.
Implementation and Decision
-5
:making
.
:

} { L, Z 1 , Z 0 , Z1 , Z 2 ,L } } { Z t , t { L, 1,0,1,2,L } { Z t

} { z1 , z2 ,L, zn1 , zn
:2 z1 , z2 ,L, zn 1 History

: 3 zn
.
:4 et = zt zt , t = 1,2,..., n zt
) ( Residuals
.
: zn +1 , zn+2 , zn +3 ,...
zn +l , l 0 zn ( 1) , zn ( 2 ) , zn ( 3) ,... zn ( l) , l 0
en ( l) = zn+l zn ( l) , l 0
:5
:6 } { z1 , z2 ,L, zn1 , zn Stationary

:
1) E ( zt ) = constant = , t
constant = 0 , t , s, t = s
2) cov ( zt , zs ) =
f ( s t ) , t , s, t s
Building Blocks

:7 White Noise Series White
{ at } Noise Process )
Independent,
( (Identically Distributed (IID 2:
1) E ( at ) = 0, t
) at : WN ( 0, 2
2 , t , s, t = s
2) cov ( at , as ) =
0 , t , s , t s
:1 : Random Walk
} { Z t:
10
Z1 = a1
Z 2 = a1 + a2
M
Z t = a1 + a2 + L + at
Z t = Z t 1 + at
a j
j Z t t
:

: ) E ( Z t ) cov ( Z t , Z s t , s
:8 Autocovariance Function :
= cov ( Z t , Z s ) , t , s
t ,s
= E ( Z t ) ( Z s ) , t , s
k Z t Z t k Z t +k
:
k = cov ( Z t , Z t k ) , k = 0, 1, 2,L
= E ( Z t ) ( Z t k ) , k = 0, 1, 2,L
:
:9 (Autocorrelation Function (ACF :
k = k , k = 0, 1, 2,L
0
:
1. 0 = 1
2. k = k
k 1
3.
:2
:
2
, k = 0
k = cov ( at , at k ) =
0, k 0
11
7 :
k 1, k = 0
=
0 0 , k 0
= k
:
Autocorrelation function of White Noise
1.0
A
utocorr
0.5
0.0
9
Lag
:10 (Partial Autocorrelation Function (PACF

Z t Z t k
Z t 1 , Z t 2 ,..., Z t k +1 k kk
kk :
Z t = k 1Z t 1 + k 2 Z t 2 + L + kk Z t k + at
: 11
Z t = 11Z t 1 + at
Z t 1
) E ( Z t 1Z t ) = 11E ( Z t 1Z t 1 ) + E ( Z t 1at
1 = 11 0
) E ( Z t 1at ) = 0 E ( Z t k at ) = 0, k = 1,2,... (
0
11 = 1
:11 :
12
k =0
k =1
k = 2,3,...
1,
1 ,
k 2 1
k 3 2
M
M
1 k
,
k 2 k 1
k 3 k 2
M
M
1
1
L
L
L
L
L
L
L
L
1
1
1
1
M
M
k 1 k 2
1
1
1
1
M
M
k 1 k 2
kk =
k
:
11 : kk
00 = 1, by definition
11 = 1
k 1
, k = 2,3,...
k k 1, j k j
j =1
k 1
1 k 1, j j
= kk
j =1
j = 1, 2,..., k 1
kj = k 1, j kkk 1,k 1 ,
: 22
11:
2 11 1 2 12
= 22
=
1 11 1
1 12
. 11 = 1
13
:3 :
11
11 = 1 = 0
1
11 kk
22 = 33 = L = 0
:
1, k = 0
kk =
0, k 0
:
Partial Autocorrelation function of White Noise
1.0
PACF
0.5
0.0
:
Lag
.
. .
9
: 12 Sample Autocorrelation Function SACF

z1 , z2 ,L, zn 1 , zn rk , k = 0,1,2,... :
, k = 0,1,2,...
1 n
zt
n t =1
) z ) ( zt + k z
2
)z
n k
( z
( z
t =1
= rk
t =1
=z
Estimator k = rk , k = 0,1,2,...
:
-1 k = 0, k > q
q
1
V ( rk ) 1 + 2 k2 , k > q
n
k =1
1
k = 0, k > 0 V ( rk ) , k > 0
n
14
-2 n k = 0 rk
:
H 0 : k = 0
H1 : k 0
:
= n rk
rk
12
= 0.05 H 0 n rk > 1.96

-3 H 0 : k = 0, k corr ( rk , rk s ) 0, s 0
-4 :
1
V ( rk ) 1 + 2 rk2 , k > q
n
k =1
:13 Sample Partial Autocorrelation

Function SPACF z1 , z2 ,L, zn 1 , zn
rkk , k = 0,1, 2,... :
k =0
1,
r,
k =1
1
1
r1 L rk 2 r1
1 L rk 3 r2
r1
M
M L M M
rkk = rk 1 rk 2 L r1 rk
,
k = 2,3,...
1
r1 L rk 2 rk 1
1 L rk 3 rk 2
r1
M
M L M
M
1
rk 1 rk 2 L r1
15
13 : rkk
r00 = 1, by definition
r11 = r1
k 1
, k = 2,3,...
rk rk 1, j rk j
j =1
k 1
1 rk 1, j rj
= rkk
j =1
j = 1,2,..., k 1
rkj = rk 1, j rkk rk 1,k 1 ,
Estimator
kk = rkk , k = 0,1, 2,...
:
1
V ( rkk ) , k > 0 -1 1
n
-2 n rkk
:
H 0 : kk = 0
H1 : kk 0
:
= n rkk
rkk
12
= 0.05 H 0 n rkk > 1.96
-3 H 0 : kk = 0, k corr ( kk ,k s ,k s ) 0, s 0
-4 :
1
V ( rkk ) , k > 0
n
:4 :
158 222 248 216 226 239 206 178 169
:
n
1
1
: zt = ( 158 + 222 + L + 169 ) = 206.89
n t =1
9
16
=z
:
, k = 0,1,2,...
n k
) z ) ( zt + k z
2
)z
( z
t =1
( z
= rk
t =1
) ( 158 222 + 222 248 + L + 178 169

= 0.265116
) ( 158 158 + 222 222 + L + 169 169
) ( 158 248 + 222 216 + L + 206 169
= r2
= -0.212
) ( 158 158 + 222 222 + L + 169 169
) ( 158 216 + 222 226 + L + 239 169
= r3
= 0.076
) ( 158 158 + 222 222 + L + 169 169
= r1
r8 = 0.230, r7 = 0.104, r6 = 0.242, r5 = 0.387, r4 = 0.183

:
q
1
V ( rk ) 1 + 2 rk2 , k > q
n
k =1
1
) V ( r1
9
1
1
2
V ( r2 ) ( 1 + 2r12 ) = 1 + 2 ( 0.265) = 0.1267
n
9
1
1
2
2
V ( r3 ) 1 + 2 ( r12 + r22 ) = 1 + 2 ( 0.265) + ( 0.212 ) = 0.1367
n
9
V ( r4 ) 0.138 V ( r5 ) 0.1454 V ( r6 ) 0.1787
: :
r00 = 1, by definition
r11 = r1 = 0.265

))
( (
k 1
, k = 2,3,...
rk rk 1, j rk j
j =1
k 1
1 rk 1, j rj
= rkk
j =1
j = 1,2,..., k 1
17
rkj = rk 1, j rkk rk 1,k 1 ,
r22 =
r2 r1, j r2 j
j =1
1
1 r1, j rj
r2 r11r1 ( 0.212 ) ( 0.265) ( 0.265) 0.282225

=
=
1 r11r1
1 ( 0.265) ( 0.265)
0.929775
j =1
= 0.30354
r21 r33
r21 = r11 r22 r11 = 0.265 ( 0.303) ( 0.265) = 0.345295
2
r33 =
r3 r2, j r3 j
j =1
k 1
1 r2, j rj
r3 ( r21r2 + r22 r1 )
1 ( r21r1 + r22 r2 )
j =1
( 0.076 ) ( ( 0.345) ( 0.212 ) + ( 0.303) ( 0.265) )

1 ( ( 0.345) ( 0.265) + ( 0.303) ( 0.212 ) )
= 0.092

r88 = 0.042, r77 = 0.013, r66 = 0.207, r55 = 0.294, r44 = 0.298
1 = 0.1111
9
:
Autocorrelation
Autocorrelation Function for Demand

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag
Corr
LBQ
Lag
Corr
LBQ
1
2
3
4
5
6
7
0.27
-0.21
-0.08
-0.18
-0.39
-0.24
0.10
0.80
-0.59
-0.21
-0.49
-1.01
-0.57
0.24
0.87
1.50
1.60
2.26
5.96
7.89
8.43
0.23
0.52
13.66
Partial Autocorrelation
Partial Autocorrelation Function for Demand

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag PAC
1
2
3
4
5
6
7
0.27
-0.30
0.09
-0.30
-0.29
-0.21
0.01
Lag PAC
0.80
-0.91
0.27
-0.89
-0.88
-0.62
0.04
8 0.04
0.13
18

_
Autoregressive-Moving Average Models
:

_ Autoregressive-Moving Average Models

.
:14 _ ) ( p, q
) ARMA ( p, q } { z1 , z2 ,K , zn1 , zn :
zt = + 1 zt 1 + 2 zt 2 + L + p zt p + at 1at 1 2 at 2 L q at q
2
) at : WN ( 0, < <
1 ,2 ,K , p Autoregressive Parameters 1 , 2 ,K , q
Moving Average Operators
Operators Algebra
:15 Backshift Operator B :
1 Bzt = zt 1
2 B m zt = B m1 ( Bzt ) = B m2 ( B ( Bzt ) ) = L = zt m
3 Bc = c, c is a constant
:
15:
-1 Forewardshift Operator F :
F = B 1
-2 Difference Operator :
) = ( 1 B
-3 Sum Operator S :
1
) S = 1 = ( 1 B
19
_ ) ( p, q
:
zt 1 zt 1 2 zt 2 L p zt p = + at 1at 1 2 at 2 L q at q
zt 1 Bzt 2 B 2 zt L p B p zt = + at 1 Bat 2 B 2 at L q B q at
L p B p ) z t = + ( 1 1 B 2 B 2 L q B q ) a t
(1 B B
2
p ( B ) z t = + q ( B ) at
2
p
p ( B ) = 1 1 B 2 B L p B Autoregressive
2
q
Operator q ( B ) = 1 1 B 2 B L q B
Moving Average Operator
:
-1 Constant Mean Model ) ARMA ( 0,0
:
0 ( B ) zt = + 0 ( B ) at
( 1) zt = + ( 1) at
) zt = + at , at : WN ( 0, 2
-2 ) ARMA ( 1,0 ) AR ( 1 :
1 ( B ) zt = + 0 ( B ) at
( 1 1 B ) zt = + at
) zt = + 1 zt 1 + at , at : WN ( 0, 2
-3 ) ARMA ( 0,1) MA ( 1 :
0 ( B ) zt = + 1 ( B ) at
z t = + ( 1 1 B ) a t
) zt = + at 1at 1 , at : WN ( 0, 2
-4 ) ARMA ( 2,0 ) AR ( 2 :
2 ( B ) zt = + 0 ( B ) at
= + at
(1 B B ) z
2
) zt = + 1 zt 1 + 2 zt 2 + at , at : WN ( 0, 2
-5 _ )1 ARMA ( 1,1) (1
:
20
1 ( B ) zt = + 1 ( B ) at
( 1 1 B ) zt = + ( 1 1B ) at
) zt = + 1 zt 1 + at 1at 1 , at : WN ( 0, 2
_ :

_
.
: :(ARMA(0,0

0 ( B ) zt = + 0 ( B ) at
zt = + at , at : WN ( 0,
)(
:
) E ( zt ) = + E ( a t
=
) at : WN ( 0, 2
) E ( zt ) = E ( zt =
:
zt = a t

zt k
E ( zt k ) ( zt ) = E ( zt k ) at
E ( zt k ) ( zt ) = k 8
k = E ( zt k ) at , k = 0, 1, 2,L
:
k = 0 : 0 = E ( zt ) at
zt = at at
E ( zt ) at = E ( at at ) = 2
2
) at : WN ( 0,
k = 0 : 0 = E ( zt ) at = 2
k = 1: 1 = E ( zt 1 ) at = 0
21
zt 1 = at 1
E ( zt 1 ) at = E ( at 1at ) = 0

zt k = at k , k = 1, 2,K
E ( zt k ) at = E ( at k at ) = 0, k = 1, 2,K
:1
, k = 0
E ( zt k ) at = E ( at k at ) =
0, k = 1,2,..
2
0 =
k = 0, k = 1, 2,K
2
:
, k = 0
k =
0, k 0
2
k 1, k = 0
=
0 0, k 0
:
Autocorrelation function of Constant Mean Model
1.0
Autocorr
k =
0 = 2
0.5
0.0
0
Lag
11
22
11 = 1 , by definition
11 = 0
1 1
1 0
2
0 0
22 = 1
=
=0
1 1
1 0
1 1
0 1
1 1 1
1 0 0
1 1 2
0 1 0
1 3
0 0 0
33 = 2
=
=0
1 1 2
1 0 0
1 1 1
0 1 0
2 1 1
0 0 1
M
1
1
1
1
M
M
k 2
kk = k 1
1
1
1
1
M
M
k 1 k 2
L
L
M
L
L
L
M
L
1
1 0 L
2
0 1 L
M
M M M
k
0 0 L
=
k 1
1 0 L
k 2
0 1 L
M M M
M
1
0 0 L
0
0
M
0 0
= = 0, k = 2,3,L
0 1
0
M
1
:
1, k = 0
kk =
0, k 0
Partial Autocorrelation function of Constant Mean Model
PACF
1.0
0.5
0.0
0
Lag

:

23
: (ARMA(1,0) = AR(1
:
1 ( B ) zt = + 0 ( B ) at
( 1 1 B ) zt = + at
) zt = + 1 zt 1 + at , at : WN ( 0, 2
)(
:
( 1 1 B ) zt = + at
1
+ ( 1 1 B ) at
) ( 1 1
1
+ E ( 1 1 B ) at
) ( 1 1
= zt
= ) E ( zt
j j
1
E ( 1 1 B ) at = E 1 B at
j =0

< Bj
j =0 1 < 1 B
Complex Variable B = a + ib
B = 1

( 1 1 B ) = 0 B > 1
1 1 B = 0
1
1
1
> 1 1 < 1
1
=B
B >1
j j
1
E ( 1 1 B ) at = E 1 B at
j =0

j j
= 1 B E ( at )
j =0
=0, t
24
E ( zt ) =
( 1 1 )
( 1 1 )
= ( 1 1 )
zt = + 1 zt 1 + at
= ( 1 1 ) + 1 zt 1 + at
= + 1 ( zt 1 ) + at
( zt ) 1 ( zt1 ) = at
zt k
E ( zt k ) ( zt ) 1E ( zt k ) ( zt 1 ) = E ( zt k ) at , k = 0, 1, 2,L
k 1 k 1 = E ( zt k ) at , k = 0, 1, 2,L
: 8
k = 0 : 0 1 1 = E ( zt ) at
:
E at ( zt ) 1E at ( zt 1 ) = E ( at at )
E at ( zt ) 1 ( 0 ) = 2
E at ( zt ) = 2
0 1 1 =
k = 1: 1 1 0 = E ( zt 1 ) at = 0
2
k 1 k 1 = 0, k = 1,2,L
k 1 k 1 = 0, k = 1, 2,L
k = 1 k 1 , k = 1, 2,L
: 0 = 1
1 = 1 0 = 1
2 = 1 1 = 1
25
k = 1k

k = , k = 0, 1, 2,L
k = k , k k
k
k = 1 , k = 0,1,2,L
:
-1 1 > 0
k
1
Autocorrelation function of AR(1) Model

0.5
0.4
0.2
ACF
0.3
0.1
0.0
10
Lag
-2 1 < 0
Autocorrelation function of AR(1) Model
0.3
0.2
0.1
0.0
-0.2
-0.3
-0.4
-0.5
10
Lag

11
26
ACF
-0.1
11 = 1 = 1 , by definition
1 1
1 1
1 2 1 12
0
22 =
=
=
=0
1 1
1 1 1 12
1 1
1 1
M
1
1
1
1
M
M
k 2
kk = k 1
1
1
1
1
M
M
k 1 k 2
1
1
1
2
1
1
M
M
M
k 1
k 2
k
1
= 1
k 1
1
1
k 2
1
1
M
M
M
k 1
k 2
1
1
1
L
L
L
L
L
L
L
L
L 1
L 12
L M
L 1k
L 1k
L 1k 1
L M
L 1
0
>0
1
:
1, k = 0
kk = 1 , k = 1
0, k 2
:
1 > 0 -1
Partial Autocorrelation function of AR(1) Model
0.5
PACF
0.4
0.3
0.2
0.1
0.0
0
10
Lag
1 < 0 -2
27
Partial Autocorrelation function of AR(1) Model

0.0
-0.1
-0.3
PACF
-0.2
-0.4
-0.5
10
Lag
: 0 = 1 00 = 1 .
:
-1
-2
-3
-4
) 1 < 1 ( ) E ( zt ) = ( 1 1 t
k t
1 1 > 0
1 < 0
) ( 00
1 1
: : (ARMA(2,0) = AR(2
:
2 ( B ) zt = + 0 ( B ) at
= + at
(1 B + B ) z
2
) zt = + 1 zt 1 + 2 zt 2 + at , at : WN ( 0, 2
:
( 1 1B 2 B ) zt = + at
2
+ ( 1 1 B 2 B 2 ) at
) ( 1 1 2
1
+ E ( 1 1B 2 B 2 ) at
) ( 1 1 2
28
= zt
= ) E ( zt
E j at j
j =0

<
2
j
j t j
j =0
j =0
2 1 < 1
2 + 1 < 1
1 < 2 < 1
)
2
( 1 1 B 2 B ) = 0 ( .
1
1
E ( 1 1 B 2 B 2 ) at = ( 1 1B 2 B 2 ) E ( at ) = 0, t
) ( 1 1 2
= ) = E ( zt
= ( 1 1 2 )
zt = ( 1 1 2 ) + 1 zt 1 + 2 zt 2 + at
= + 1 ( zt 1 ) + 2 ( zt 2 ) + at
( zt ) 1 ( zt1 ) 2 ( zt2 ) = at
zt k :
) 1 ( zt 1 ) ( zt k ) 2 ( zt 2 ) ( zt k )
E ( zt ) ( zt k
= E at ( zt k ) , k = 0, 1, 2,...
E ( zt ) ( zt k ) 1E ( zt 1 ) ( zt k ) 2 E ( zt 2 ) ( zt k )
= E at ( zt k ) , k = 0, 1, 2,...
k 1 k 1 2 k 2 = E at ( zt k ) , k = 0, 1, 2,...
8 :
k = 0 : 0 1 1 2 2 = E at ( zt ) = 2 0 = 1 1 2 2 + 2
29
k = 1: 1 1 0 2 1 = 0 1 = 1 0 2 1
k = 2 : 2 1 1 2 0 = 0 2 = 1 1 2 0
k 1: k = 1 k 1 + 2 k 2
k = 1 k 1 + 2 k 2 , k = 1,2,...
) : k 1 k 1 2 k 2 = 0, k = 1, 2,...

(
:
1 0 = 1
2 1 = 1 0 + 2 1 1 = 1
1 2

2
2 = 1 1 + 2 0 2 = 1 + 2
1 2

) AR ( 2
-1 )1 = 0.4, 2 = 0.4 (1
)1 = 1.5, 2 = 0.8 (2
-2
)1 = 0.5, 2 = 0.6 (3
-3
)(1
ACF
0.7
0.6
0.5
0.3
ACF
0.4
0.2
0.1
0.0
20
10
Lag
ACF
1.0
)(2
0.5
ACF
0.0
30
20
10
Lag
-0.5
0
)(3
ACF
0.5
ACF
0.0
-0.5
20
10
Lag
) AR ( 2:
1 1
1 2 2 12
= 22
=
0
1 1
1 12
1 1
1 1 1
1 1
1 = 1 + 2 1
1 1 2
1 1
2 = 1 1 + 2
1 3
1 3 = 1 2 + 2 1
33 = 2
= 2
=0
1 1 2
>0
1 1 1
2 1 1

31
1
1
1
1
M
M
k 2
kk = k 1
1
1
1
1
M
M
k 1 k 2
1
1
2
1
M
M
k
= k 1
k 1
k 2
M
1
L
L
L
L
L
L
L
L
1
1
M
k 2
L
1 = 1 0 + 2 1
L
2 = 1 1 + 2 0
L
M
L k = 1 k 1 + 2 k 2
= 0, k = 3,4,...
>0
k =0
1,
,
k =1
1
kk = 2 12
1 2 , k = 2
1
k 3
0,
AR ( 2 )
1 = 0.4, 2 = 0.4 (4) -4
1 = 1.5, 2 = 0.8 (5)
-5
1 = 0.5, 2 = 0.6 (6)
-6
(4)
PACF
0.7
0.6
0.4
0.3
0.2
0.1
0.0
0
10
20
Lag
PACF
1
PACF
PACF
0.5
-1
0
10
Lag
32
20
(5)
(6)
PACF
0.3
0.2
PACF
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
0
10
20
Lag
: (ARMA(0,1) = MA(1 :
:
0 ( B ) zt = + 1 ( B ) at
z t = + ( 1 1 B ) a t
zt = + at 1at 1 , at : WN ( 0, 2 )
:
E ( zt ) = E ( + at 1at 1 ) =
=

zt = at 1at 1
zt k
E ( zt ) ( zt k ) = E ( zt k ) at 1 E ( zt k ) at 1 , k = 0, 1, 2,...
k = E ( zt k ) at 1E ( zt k ) at 1 , k = 0, 1, 2,...
k = 0 : 0 = E ( zt ) at 1 E ( zt ) at 1
: E ( zt ) at 1 E ( zt ) at
E ( zt ) at = E ( at at ) 1E ( at 1at ) = 2
E ( zt ) at 1 = E ( at at 1 ) 1 E ( at 1at 1 ) = 1 2
0 = 2 1 ( 1 2 ) = 2 ( 1 + 12 )
33
k = 1: 1 = E ( zt 1 ) at 1E ( zt 1 ) at 1
1
1
=
0 1 + 12
= 1 = 1 2 1
k = 2 : 2 = E ( zt 2 ) at 1E ( zt 2 ) at 1
2 = 0 2 = 0
k 2 : k = 0 k = 0
) MA ( 1
:
1,
k =0
1
k =
, k =1
2
1
+
0
k 2
:
-1 1 = 0.8
ACF
0.0
-0.1
ACF
-0.2
-0.3
-0.4
-0.5
20
10
Lag
-2 1 = 0.8
ACF
0.5
0.4
0.2
0.1
0.0
20
10
Lag
34
ACF
0.3
MA ( 1)
1 1
1 1
12 ( 1 12 )
1 2
1 0
12
12
22 =
=
=
=
=
1 1
1 12
1 12 1 + 12 + 14
1 16
1 1
1
1
33 = 2
1
1
2
1
1
2
1
3
0
=
2
1
1
1
1
0
1
1
1
1
1
1
1
0
13 ( 1 12 )
0
13
=
=
0 1 2 12
1 18
1
1

( 1 ) ,
2
1
k >0
1 1 (
2 k +1)
:
1 = 0.8 -1
PACF
0.5
0.4
PACF
0.3
0.2
0.1
0.0
-0.1
1 = 0.8 -2
-0.2
-0.3
0
10
20
Lag
PACF
0.0
-0.1
PACF
kk =
k
1
1
1
1
1
1
1
-0.2
-0.3
-0.4
-0.5
0
35 10
Lag
20
: )ARMA(0,2
: (= MA(2
:
0 ( B ) zt = + 2 ( B ) at
z t = + ( 1 1 B 2 B 2 ) a t
) zt = + at 1at 1 2 at 2 , at : WN ( 0, 2
:
E ( zt ) = E ( + at 1at 1 2 at 2 ) =
=

zt k
) at 1 E ( zt k ) at 1
) at 2 , k = 0, 1, 2,...

zt = at 1at 1 2 at 2
E ( zt ) ( zt k ) = E ( zt k
2 E ( zt k ) at 2 , k = 0, 1, 2,...
k = E ( zt k ) at 1E ( zt k ) at 1 2 E ( zt k
0 = ( 1 + 12 + 22 ) 2
1 = ( 1 + 1 2 ) 2
2 = 2 2
k = 0, k > 2
1 + 1 2
1 + 12 + 22
= 1
2
1 + 12 + 22
= 2
k = 0, k > 2

36
1,
k =0
+
1 2 1 22 , k = 1
1 + 1 + 2
k =
2
, k =2
1 + 12 + 22
0,
k >2
) MA ( 2
-7 )1 = 0.4, 2 = 0.4 (7
)1 = 1.5, 2 = 0.8 (8
-8
)1 = 0.5, 2 = 0.6 (9
-9
)(7
ACF
0.0
-0.1
ACF
-0.2
-0.3
20
10
Lag
)(8
ACF
0.2
0.1
0.0
-0.1
-0.3
-0.4
-0.5
-0.6
-0.7
20
10
Lag
)(9
ACF
0.4
0.3
0.2
0.1
-0.2
-0.3
-0.4
20
10
Lag
37
-0.5
0
ACF
0.0
-0.1
ACF
-0.2

) MA ( 2 11
:
)1 = 0.4, 2 = 0.4 (10
-10
)1 = 1.5, 2 = 0.8 (11
-11
)1 = 0.5, 2 = 0.6 (12
-12
)(10
PACF
0.0
-0.1
PACF
-0.2
-0.3
20
10
Lag
)(11
PACF
0.2
0.1
0.0
-0.1
-0.3
PACF
-0.2
-0.4
-0.5
-0.6
-0.7
20
10
Lag
)(12
PACF
0.4
0.3
0.2
0.1
-0.2
-0.3
-0.4
-0.5
10
20
Lag
38
PACF
0.0
-0.1
: -
: (ARMA(1,1
:
1 ( B ) zt = + 1 ( B ) at
( 1 1B ) zt = + ( 1 1B ) at
zt 1 zt 1 = + at 1at 1
zt = + 1 zt 1 + at 1at 1 , at : WN ( 0, 2 ) , 1 1
1 < 1 1 < 1
Degeneracy Condition 1 1
1 = 1 ( 1 1B ) zt = + ( 1 1B ) at
) ( 1 1B zt = + at
1 1
= ) ARMA ( 0,0
( 1 1B ) zt = + ( 1 1B ) at
( 1 1 B ) a
= zt
+
t
) 1 1 ( 1 1B
( 1 1 B ) E a
= ) E ( zt
+
)( t
) 1 1 ( 1 1B
1 < 1
1 1
= ) E ( zt
E ( zt ) = = 1 ) = ( 1 1
1
zt = ( 1 1 ) + 1 zt 1 + at 1at 1
( zt ) 1 ( zt 1 ) = at 1at 1
( zt k ) , k = 0, 1, 2,...
E ( zt k ) ( zt ) 1E ( zt k ) ( zt 1 ) = E ( zt k ) at 1E ( zt k ) at 1 ,
k = 0, 1, 2,...
k 1 k 1 = E ( zt k ) at 1E ( zt k ) at 1 , k = 0, 1, 2,...
k = 0 0 1 1 = E ( zt ) at 1E ( zt ) at 1
E ( zt ) at E ( zt ) at 1
( zt ) 1 ( zt 1 ) = at 1at 1
at at 1
] E ( zt ) at 1E ( zt 1 ) at = E [ at at ] 1E [ at 1at
39
E ( zt ) at 1 ( 0 ) = 1 ( 0 )
2
E ( zt ) at = 2
E ( zt ) at 1 1 E ( zt 1 ) at 1 = E [ at at 1 ] 1E [ at 1at 1 ]
E ( zt ) at 1 1 2 = 0 1 2
E ( zt ) at 1 = 2 ( 1 1 )

k = 0 0 1 1 = 2 1 2 ( 1 1 )
0 1 1 = 2 1 1 ( 1 1 )
k = 1 1 1 0 = E ( zt 1 ) at 1E ( zt 1 ) at 1
1 1 0 = 1 2
k = 2 2 1 1 = E ( zt 2 ) at 1E ( zt 2 ) at 1 = 0
k 2 k 1 k 1 = 0
0 1 1 = 1 1 ( 1 1 )
2
1 1 0 = 1
0 =
1 =
1 =
1 + 211 2
1 12
2
1
( 1 11 ) ( 1 1 ) 2
1 12
1 ( 1 11 ) ( 1 1 )
=
0
1 + 12 211
k 1 k 1 = 0, k 2
0
k 1 k 1 = 0, k 2
0 = 1 k 2
( 1 11 ) ( 1 1 )
1 =
1 + 12 211
40
2 = 1 1
) ( 1 11 ) ( 1 1
2 = 1
1 + 12 211
3 = 1 2
) ( 1 11 ) ( 1 1
1 + 12 211
3 = 12
.
) ARMA ( 1,1
13 1 = 0.9,1 = 0.5
1,
k =0
) ( 1 11 ) ( 1 1
k =
, k =1
2
1
+
1
1
1
1 k 1
k 2
)(13
)ACF of ARMA(1,1
1.0
0.9
0.8
0.7
0.6
C1
0.5
0.4
0.3
0.2
0.1
0.0
15
10
Lag
14 1 = 0.9,1 = 0.5
)(14
)ACF of ARMA(1,1
0.5
-0.5
15
10
Lag
41
C1
0.0
) ARMA ( 1,1
) AR ( 1
) 1 ( k = 1k 11 , k 2
kk 11 11 :
11 kk
) ( 1 11 ) ( 1 1
1 + 12 211
= 11 = 1
2 11 1
1 11 1

33 = 3 21 2 22 1 , 21 = 11 2211
1 21 1 22 2
= 22
.
1 = 0.9,1 = 0.5
11 = 0.944186 22 = -0.384471 33 = 0.183710

44 = -0.908462 55 = 0.452979 66 = -0.226337
77 = 0.113154 88 = -0.565702 99 = 0.282834
15
15
)PACF of ARMA(1,1
1.0
0.5
C2
0.0
15
10
Lag
16 1 = 0.9,1 = 0.5
16
)PACF of ARMA(1,1
0.3
0.2
0.1
0.0
-0.2
-0.3
-0.4
-0.5
-0.6
15
10
Lag
42
C2
-0.1
) ARMA ( 1,1
)MA ( 1
. 11 = 1
: (ARMA(p,q
: (AR(p
:
-1 .
-2 k > p
11 = 22 = 33 = L = pp 0
p +1, p +1 = p +2, p +2 = L = 0
. k > p
: : (MA(q
:
k
>
q
-1
1 = 2 = 3 = L = q 0
q +1,q+1 = q+ 2,q +2 = L = 0
. k > q
-2
.
Duality AR .MA
: :(ARMA(p,q
:

. k
k > q p
k > p q .
43

: Nonstationar Time Series Models
: :
6
E ( zt ) = = constant t

) zt = b0 + b1t + at , at : WN ( 0, 2 ) , b0 , b1 ( ,
E ( z ) t = b0 + b1t
.
zt
wt = zt = zt zt 1 = b0 + b1t + at b0 b1 ( t 1) at 1
=b1 + at at 1 = b1 + ct
) wt = b1 + ct , ct : WN ( 0, 2
) : 2 ( 2
wt
E ( wt ) = b1 = constant t
) = ( 1 B )
( .

) zt = b0 + b1t + b2t 2 + at , at : WN ( 0, 2 ) , b0 , b1 , b2 ( ,
E ( zt ) = b0 + b1t + b2t
. ) 2 zt (
) + b1t + b2t 2 + at
) 2 zt = 2 ( b0 + b1t + b2t 2 + at
( 1 2B + B ) z = ( 1 2B + B ) ( b
2
wt = { b0 2b0 + b0 } + { b1t 2b1 ( t 1) + b1 ( t 2 ) } +
}+
) 2b2 ( t 1) + b2 ( t 2
2
{b t
2
} { at 2at 1 + at 2
} = 2b2 + { at 2at 1 + at 2
) =b + ht , ht : WN ( 0, 2
44
) wt = 2 zt = b + ht , ht : WN ( 0, 2
E ( wt ) = b = constant t
) 2 (
.
2
2
) : .(

) zt = b0 + b1t + L + bd t + at , at : WN ( 0, 2 ) , b0 , b1 ,L , bd ( ,
d
d zt wt = d zt .
:16
) ) , b , b ,L, b ( ,
d
zt = b0 + b1t + L + bd t + at , at : WN ( 0,
d
d zt d .
: : :
6
. t

V ( zt ) = 0 = constant t
) zt = zt 1 + at , at : WN ( 0, 2
zt = a1 + a2 + L + at
E ( zt ) = 0 = constant t
V ( zt ) = t 2
. t

wt = zt = zt zt 1 = at

E ( wt ) = 0 = constant t
V ( wt ) = 2 = constant t
.
)(
) V ( zt ) = cf ( t
c > 0 ) ( f t
) T ( zt
) ( T .
45
z 1
= ) y t = T ( zt
) ( , .
:
1.0
zt
0.5
0.0
-0.5
zt
ln zt
1
zt
yt
-0.1
1
zt
:
)( zt
Original Series
400
300
)z(t
200
100
90
80
70
60
50
40
30
20
10
Index
)( yt = ln zt
Transformed Series
6.0
5.5
)lnz(t
5.0
90
80
70
60
50
40
30
20
)( yt yt = yt yt 1
46
10
Index
Differenced and Transformed Series

0.2
0.1
)y(t)-y(t-1
0.0
-0.1
-0.2
90
80
70
60
50
40
30
20
Index
10
-- ))p,d,q
Autoregressive-Integrated-Moving Average Models
(ARIMA(p,d,q
wt = d zt -
) ( p, q:
) p ( B ) wt = p ( B ) d zt = + q ( B ) at , at : WN ( 0, 2
p ( B ) ( 1 B ) zt = + q ( B ) at , at : WN ( 0,
d
-- ) ( p, d , q
) ( , .
-- ) : ( p, d , q
: - ) (1,1 : (ARIMA(1,1,0) =ARI(1,1

) 1 ( B ) ( 1 B ) zt = + 0 ( B ) at , at : WN ( 0, 2
( 1 1B ) ( 1 B ) zt = + at
{ 1 ( 1 + 1) B + 1B 2 } zt = + at
1 < 1
),
zt = + ( 1 + 1) zt 1 1 zt 2 + at , at : WN ( 0,
: - )(1,1
: (ARIMA(0,1,1) = IMA(1,1

) 0 ( B ) ( 1 B ) zt = + 1 ( B ) at , at : WN ( 0, 2
1 < 1
at : WN ( 0, 2 ) ,
( 1 B ) zt = + ( 1 1B ) at ,
zt zt 1 = + at 1at , at : WN ( 0, 2 ) , 1 < 1
47
zt = + zt 1 + at 1at , at : WN ( 0, 2 ) , 1 < 1
: Random Walk with Trend Model

: (ARIMA(0,1,0

) 0 ( B ) ( 1 B ) zt = + 0 ( B ) at , at : WN ( 0, 2
) at : WN ( 0, 2
( 1 B ) z t = + at ,
zt = + zt 1 + at , at : WN ( 0,
) ( B :(ARMA(p,q
) ARMA ( p, q
) p ( B ) zt = + q ( B ) at , at : WN ( 0, 2
p ( B ) ( zt ) = q ( B ) at , at : WN ( 0,
) p ( B
) q ( B
) at , at : WN ( 0, 2
)p ( B
) p ( 1
) q ( B
) at , at : WN ( 0, 2
)p ( B
= zt
= zt
) q ( B

)p ( B
( 1) =
p
p ( B ) = 0
) q ( B
) at , at : WN ( 0, 2
)p ( B
) q ( B
)p ( B
) q ( B
= 0 B 0 + 1B 1 + 2 B 2 + 3 B 3 + L , 0 = 1
)p ( B
48
= zt
= ) ( B
= ) ( B
:17 ) ARMA ( p, q
) q ( B
= 0 B 0 + 1B 1 + 2 B 2 + 3 B 3 + L , 0 = 1
)p ( B
= ) ( B
) q ( B
= j B j , 0 = 1
p ( B ) j =0
= ) ( B
0 = 1, 1 , 2 , 3 ,L
) ( B
q
2
: ) zt = B at , at : WN ( 0,
)
(p
) . ARMA ( p, q
:
: (AR(1
(AR(1
) 1 ( B ) ( zt ) = 0 ( B ) at , at : WN ( 0, 2
( 1 1B ) ( zt ) = at
1
a
( 1 1B ) t
= zt
zt = ( B ) a t
1
) ( 1 1B
= ) ( B
1 , 2 , 3 ,L
1
) ( 1 1B
= ) ( B
( B ) ( 1 1B ) 1
B 2 + 3 B 3 + L) ( 1 1B ) 1
( 1 + B +
1
B j .
B j
1 < 1
B 2 + 3 B 3 + L) ( 1 1B ) 1,
( 1 + B +
1
B 0 : ( 1) ( 1) 1
B1 : 1 1 0 1 = 1
B 2 : 2 11 0 2 = 11 = 12
B 3 : 3 21 0 3 = 21 = 13
M
B j : j j 11 0 j = j 11 = 1j
(AR(1
49
j = 1j , 1 < 1
: (MA(1
( MA(1
0 ( B ) ( zt ) = 1 ( B ) at , at : WN ( 0, 2 )
( zt ) = ( 1 1B ) at
zt = ( B ) a t
( B ) = ( 1 1 B )
B j
1 = 1 , 2 = 3 = L = 0
j=0
j =1
j2
1,
j = 1 ,
0,
: (AR(2
( AR(2
2 ( B ) ( zt ) = 0 ( B ) at , at : WN ( 0, 2 )
(1 B B ) ( z ) = a
2
zt =
1
a
( 1 1B 2 B 2 ) t
zt = ( B ) a t
( B) =
1
( 1 1B 2 B 2 )
( B ) ( 1 1B 2 B 2 ) 1
( 1 + B +
1
1 , 2 , 3 ,L
B 2 + 3 B 3 + L) ( 1 1B 2 B 2 ) 1
B1 : 1 1 = 0 1 = 1
2
B 2 : 2
1 1 2 = 0 2 =
1 1 + 2 = 1 + 2
B 3 : 3
1 2 2 1 = 0 3 =
1 2 + 2 1
M
B j : j
1 j 1 2 j 2 = 0 j =
1 j 1 + 2 j 2
( AR(2
1,
,
1
j = 2
1 + 2 ,
1 j 1 + 2 j 2 ,
j=0
j =1
j=2
j3
50
: (MA(2
( MA(2
0 ( B ) ( zt ) = 2 ( B ) at , at : WN ( 0, 2 )
( zt ) = ( 1 1B 2 B 2 ) at
zt = ( B ) a t
( B ) = ( 1 1 B 2 B
B j
1 = 1 , 2 = 2 , 3 = 4 = 5 L = 0
j=0
j =1
j=2
j2
1,
,
j = 1
2 ,
0,
: (ARMA(1,1
( ARMA(1,1
1 ( B ) ( zt ) = 1 ( B ) at , at : WN ( 0, 2 )
( 1 1B ) ( zt ) = ( 1 1B ) at
( 1 1 B ) a
zt =
( 1 1B ) t
zt = ( B ) a t
( B) =
( 1 1 B )
( 1 1B )
1 , 2 , 3 ,L
( B ) ( 1 1B ) ( 1 1 B )
( 1 + B +
1
B 2 + 3 B 3 + L) ( 1 1B ) ( 1 1B )
B1 : 1 1 = 1 1 = 1 1
B 2 : 2
1 1 = 0 2 =
1 1 = 1 ( 1 1 )
2
B 3 : 3
1 2 = 0 3 =
1 2 = 1 ( 1 1 )
M
j 1
B j : j
( 1 1 )
1 j 1 = 0 j =
1 j 1 = 1
j =
1 j 1 =
j 1
1
( 1 1 ) ,
( ARMA(1,1
j 1,
1 < 1, 1 1
: (ARI(1
( ARI(1
51
1 ( B ) ( 1 B ) ( zt ) = at , at : WN ( 0, 2 )
zt =
1
a
( 1 1B ) ( 1 B ) t
zt = ( B ) a t
( B) =
1
( 1 1B ) ( 1 B )
1 , 2 , 3 ,L
( B ) ( 1 1B ) ( 1 B ) 1
( 1 + B + B
( 1 + B + B
1
2
2
+ 3 B 3 + L) ( 1 1B ) ( 1 B ) 1
+ 3 B 3 + L) ( 1 ( 1 + 1) B + 1B 2 ) 1
B1 : 1 ( 1 + 1) = 0 1 = 1 + 1
B 2 : 2 ( 1 + 1) 1 + 1 = 0 2 = ( 1 + 1) 1 + 1 = ( 1 + 1) + 1
2
B 3 : 3 ( 1 + 1) 2 +
1 1 = 0 3 = ( 1 + 1) 2
1 1
M
B j : j ( 1 + 1) j 1 +
1 j 2 = 0 j = ( 1 + 1) j 1
1 j 2
( ARI(1
j=0
j =1
1,
+ 1,
1
j =
2
( 1 + 1) + 1 ,
( 1 + 1) j 1
1 j 2 ,
j=2
j3

(ARIMA(1,0,1 Random Walk Mdel

zt = zt 1 + at , at : WN ( 0, 2 )
zt zt 1 = at , at : WN ( 0,
( 1 B ) z t = at
zt =
1
a
(1 B) t
( B) =
1
( 1 B)
1 , 2 , 3 ,L
52
( B) ( 1 B) 1
B 2 + 3 B 3 + L) ( 1 B ) 1
( 1 + B +
1
B1 : 1 1 = 0 1 = 1
B2 : 2 1 = 0 2 = 1 = 1
B3 : 3 2 = 0 3 = 2 = 1
M
B j : j j 1 = 0 j = j 1 = 1
) ARIMA ( 0,1,0
j = 1,
j 1
) : ( B
- (ARMA(p,q
) zt = ( B ) at , at : WN ( 0, 2
zt = at + 1at 1 + 2at 2 + 3at 3 + L
= j at j , 0 = 1
j =0
< 2j
j =0
:1
- (ARMA(p,q
< zt = j at j , at : WN ( 0, 2 ) , 0 = 1, j =0 2j
j =0
-1
E ( z t ) = , t
-2
, k = 0,1,2,L
j +k
2
j
j =0
= k
j =0
: (AR(1
j = , 1 < 1
j
1
53
1k
1 12
=
= 1k , k = 0,1, 2,L
1
1 12
1j1j+k
j =0
2j
1
j =0
j j+k
j =0
2
j

: (2) 2
(AR(2), MA(1), MA(2), ARMA(1,1), ARMA(2,1), ARMA(1,2
54
j =0
= k

(ARMA(p,q
Minimum Mean Square Error Forecasts for
ARMA(p,q) Models
-
(ARMA(p,q
< zt = j at j , at : WN ( 0, 2 ) , 0 = 1, j =0 2j
j =0
zt = at + 1at 1 + 2at 2 + 3at 3 + L
= j at j , 0 = 1
j =0
: (ARIMA(p,d,q .
} { z1 , z2 ,L, zn 1 , zn zn ( l) , l 1
zn +l, l 1
zn ( l) = 0an + 1an 1 + 2an 2 + L , l 1
zn +l, l 1
zn + l = an + l + 1an + l1 + L + l1an +1 + lan + l+1an 1 + L , l 1
) ( 5
2
E zn + l zn ( l) = E an + l + 1an + l1 + L + l1an +1 + ( l 0 ) an + ( l+1 1 ) an 1 + L

2
= ( 1 + 12 + L + l21 ) 2 + ( l+ j j ) 2
2
j =0

j j j
j = 0,1, 2,L , l 1
j = l+ j ,
MMSE Forecasts

zn ( l) = lan + l+1an 1 + l+ 2an 2 + L , l 1
:2
:
en ( l) = zn + l zn ( l) = an +l + 1an +l1 + 2an +l2 + L + l1an +1 , l 1
V en ( l) = 2 ( 1 + 12 + 22 + L + l21 ) , l 1
55
zn ( l) = lan + l+1an 1 + l+ 2an 2 + L, l 1

zn +l, l 1
} . { a1 , a2 ,L, an 1 , an
: 18 ) } I ( { z1 , z2 ,L, zn 1, zn
) } I ( { a1 , a2 ,L, an 1 , an } { a1 , a2 ,L, an 1 , an
} . { z1 , z2 ,L , zn 1, zn
: } { z1 , z2 ,L, zn 1 , zn
} { a1 , a2 ,L, an 1 , an .
: 3 MMSE Forecasts
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
zn +l, l 1 } . { z1 , z2 ,L , zn 1, zn
2
zn ( l) = lan + l+1an 1 + l+ 2an 2 + L , l 1
.
:2
a , j 0
1 E ( an + j zn , zn 1 ,L) = n + j
j>0
0,
j0
zn + j ,
2 E ( zn + j zn , zn 1 ,L) =
zn ( j ) , j > 0
3 2
zn + l, l 1
:19
zn ( l) , l 1 l 1 n .
: (ARIMA(p,d,q
: : (AR(1
} { z1 , z2 ,L , zn 1, zn n
(AR(1
56
zt = 1 ( zt 1 ) + at , at : WN ( 0, 2 ) , 1 < 1, ( , )
. zn +l, l 1 zn +1 , zn +2 , zn +3 ,L
3
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
= +E 1 ( zn + l1 ) + an +l zn , zn 1 ,L , l 1
= +E 1 ( zn + l1 ) zn , zn 1 ,L + an +l zn , zn 1 ,L , l 1
= +1E ( zn + l1 zn , zn1 ,L) + E an + l zn , zn 1 ,L , l 1
zn ( l) = +1E ( zn + l1 zn , zn 1 ,L) + E an + l zn , zn 1 ,L , l 1
l = 1: zn ( 1) = +1E ( zn zn , zn 1 ,L) + E an +1 zn , zn 1,L

= +1 ( zn )
l = 2 : zn ( 2 ) = +1E ( zn +1 zn , zn 1 ,L) + E an + 2 zn , zn 1,L

= +1 zn ( 1)
l = 3 : zn ( 3) = +1E ( zn + 2 zn , zn 1 ,L) + E an + 3 zn , zn 1 ,L
= +1 zn ( 2 )
zn ( l) = +1 zn ( l 1) , l 1
(AR(1
:20
l = 1 Continuity Condition
zn ( l 1) = zn ( 0 ) = zn
2
V en ( l) = 2 ( 1 + 12 + 22 + L + l21 ) , l 1
( AR(1
j = 1j , 1 < 1
V en ( l) = 2 1 + 12 + 14 + L + 1 (
=2
2 l1)
) , l 1
1 12 l
, l 1
1 12
57
zt 0.97 = 0.85 ( zt 1 0.97 ) + at , at : WN ( 0,0.024 )
z156 = 0.49
. z157 , z158 , z159
zn ( l) = +1 zn ( l 1) , l 1 :
z156 ( 1) = 0.97+0.85 ( z156 0.97 )
= 0.97+0.85 ( 0.49 0.97 ) = 0.56
z156 ( 2 ) = 0.97+0.85 ( z156 ( 1) 0.97 )
= 0.97+0.85 ( 0.56 0.97 ) = 0.62
z156 ( 3) = 0.97+0.85 ( z156 ( 2 ) 0.97 )
= 0.97+0.85 ( 0.62 0.97 ) = 0.68
V en ( l) = 2
1 12 l
, l 1
1 12
V e156 ( 1) = 0.024
V e156 ( 2 ) = 0.024
V e156 ( 2 ) = 0.024
1 ( 0.85)
1 ( 0.85)
1 ( 0.85)
1 ( 0.85)
= 0.041
= 0.054
: (AR(2 :
n { z1 , z2 ,L , zn 1, zn }
( AR(2
zt = + 1 ( zt 1 ) + 2 ( zt 2 ) + at , at : WN ( 0, 2 ) , ( , ) ,
2 1 < 1, 2 + 1 < 1, 2 < 1

. zn +l, l 1 zn +1 , zn +2 , zn +3 ,L
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
= +E 1 ( zn + l1 ) + 2 ( zn + l2 ) + an + l zn , zn 1 ,L , l 1
= +E 1 ( zn + l1 ) zn , zn 1 ,L + 2 ( zn + l2 ) zn , zn1 ,L + an + l zn , zn 1 ,L , l 1
= +1E ( zn + l1 zn , zn1 ,L) + 2 E ( zn + l 2 zn , zn 1 ,L) + E an + l zn , zn 1,L , l 1
zn ( l) = +1E ( zn + l1 zn , zn 1 ,L) + 2 E ( zn + l2 zn , zn 1 ,L) + E an + l zn , zn 1 ,L , l 1
2
58
l = 1: zn ( 1) = +1E ( zn zn , zn 1 ,L) + 2 E ( zn 1 zn , zn 1,L) + E an +1 zn , zn 1 ,L

= +1 ( zn ) + 2 ( zn 1 )
l = 2 : zn ( 2 ) = +1E ( zn +1 zn , zn 1 ,L) + 2 E ( zn zn , zn 1 ,L) + E an + 2 zn , zn 1 ,L
= +1 zn ( 1) + 2 ( zn )
l = 3 : zn ( 3) = +1E ( zn + 2 zn , zn 1 ,L) + 2 E ( zn +1 zn , zn 1,L) + E an +3 zn , zn 1 ,L

= +1 zn ( 2 ) +2 zn ( 1)
l = 4 : zn ( 4 ) = +1E ( zn +3 zn , zn 1 ,L) + 2 E ( zn + 2 zn , zn 1,L) + E an + 4 zn , zn 1 ,L

= +1 zn ( 3) +2 zn ( 2 )
zn ( l) = +1 zn ( l 1) + 2 zn ( l 2 ) , l 1
(AR(2
2
1,
,
1
j = 2
1 + 2 ,
1 j 1 + 2 j 2 ,
j=0
j =1
j=2
j3
: (ARIMA(0,1,1 :
n { z1 , z2 ,L , zn 1, zn }
( ARIMA(0,1,1
zt = zt 1 + at 1at 1 , at : WN ( 0, 2 )
. zn +l, l 1 zn +1 , zn +2 , zn +3 ,L
3
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
=E ( zn + l1 zn , zn 1 ,L) + E ( an +l zn , zn 1 ,L) 1E ( an+ l1 zn , zn 1 ,L) , l 1
zn ( l) =E ( zn + l1 zn , zn 1 ,L) + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) , l 1
l = 1: zn ( 1) =E ( zn zn , zn 1 ,L) + E ( an+1 zn , zn 1 ,L) 1E ( an zn , zn1 ,L)

= zn 1an
l = 2 : zn ( 2 ) =E ( zn +1 zn , zn 1 ,L) + E ( an + 2 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L)

= zn ( 1)
l = 3 : zn ( 3) =E ( zn + 2 zn , zn 1 ,L) + E ( an +3 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L)

= zn ( 2 )
zn ( l) = zn ( l 1) , l 2
59

(ARIMA(0,1,1
zn 1an , l = 1
zn ( l) =
zn ( l 1) , l > 1
: : (MA(1
} { z1 , z2 ,L , zn 1, zn n
(MA(1
) zt = + at 1at 1 , at : WN ( 0, 2
zn +1 , zn +2 , zn +3 ,L . zn +l, l 1
3
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
= + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) , l 1
zn ( l) = + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) , l 1
)l = 1: zn ( 1) = + E ( an +1 zn , zn 1 ,L) 1E ( an zn , zn 1 ,L
= 1an
)l = 2 : zn ( 2 ) = + E ( an + 2 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L
=
)l = 3 : zn ( 3) = + E ( an +3 zn , zn 1 ,L) 1E ( an + 2 zn , zn 1 ,L
=
zn ( l) = , l 2
(MA(1

1an , l = 1
zn ( l) =
l 2
,
: : (MA(2
} { z1 , z2 ,L , zn 1, zn n
(MA(2
) zt = + at 1at 1 2at 2 , at : WN ( 0, 2
zn +1 , zn +2 , zn +3 ,L . zn +l, l 1
3
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
= + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) 2 E ( an +l2 zn , zn 1,L) , l 1
60
zn ( l) = + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) 2 E ( an + l2 zn , zn 1 ,L) , l 1
l = 1: zn ( 1) = + E ( an +1 zn , zn 1 ,L) 1E ( an zn , zn 1 ,L) 2 E ( an 1 zn , zn 1 ,L)

= 1an 2an 1
l = 2 : zn ( 2 ) = + E ( an + 2 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L) 2 E ( an zn , zn 1 ,L)

= 2 an
l = 3 : zn ( 3) = + E ( an +3 zn , zn 1 ,L) 1E ( an + 2 zn , zn 1 ,L) 2 E ( an +1 zn , zn 1 ,L)

=
zn ( l) = , l 3
(MA(2

1an 2an 1 , l = 1
zn ( l) = 2an ,
l= 2
,
l 3
: (ARMA(1,1 :
n { z1 , z2 ,L , zn 1, zn }
( ARMA(1,1
zt = + 1 ( zt 1 ) + at 1at 1 , at : WN ( 0, 2 ) , 1 1 , 1 < 1
. zn +l, l 1 zn +1 , zn +2 , zn +3 ,L
3
zn ( l) = E ( zn +l zn , zn 1 ,L) , l 1
= + 1E ( zn + l1 ) zn , zn 1 ,L + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) , l 1
zn ( l) = + 1E ( zn + l1 ) zn , zn 1 ,L + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) , l 1
l = 1: zn ( 1) = + 1E ( zn ) zn , zn 1 ,L + E ( an +1 zn , zn1 ,L) 1E ( an zn , zn 1,L)

= + 1 ( zn ) 1an
l = 2 : zn ( 2 ) = + 1E ( zn +1 ) zn , zn 1 ,L + E ( an + 2 zn , zn1 ,L) 1E ( an+1 zn , zn 1 ,L)
= + 1 zn ( 1)
l = 3 : zn ( 3) = + 1E ( zn + 2 ) zn , zn 1 ,L + E ( an +3 zn , zn 1,L) 1E ( an + 2 zn , zn 1 ,L)
= + 1 zn ( 2 )
zn ( l) = + 1 zn ( l 1) , l 2

( ARMA(1,1
61
l=1
l 2
+ 1 ( zn ) 1an ,
zn ( l) =
+ 1 zn ( l 1) ,
:
(ARMA(1,1
zt = + 1 ( zt 1 ) + at 1at 1 , at : WN ( 0, 2 ) , 1 1 , 1 < 1
1 1 ) ARMA(1,1) IMA(1,1
. (IMA(1,1
:
:
ARIMA(1,1,1), ARIMA(2,1,0), ARIMA(0,1,2), ARIMA(1,2,0),
.(ARIMA(0,2,1), ARIMA(0,2,0
: Forecasting Limits
zn ( l) , l 1 Point
Forecast

P ( Z n +m = zn ( m ) ) = 0, for some m > 0
) (

.

Interval Forecast ] [ a, b

) P ( a Z n+m b ) = ( 1

a b ) 1 ( 100 ( 1 ) %
= 0.05 %95
a . b
2
:21 ) at : N ( 0, 100 ( 1 ) %
zn +l, l 1
12
zn ( l) u 2 V en ( l)

u 2 100 1 ). N ( 0,1
2
= 0.05 . u0.025 = 1.96

: ) at : N ( 0, 2
.
62
:
) zt 0.97 = 0.85 ( zt 1 0.97 ) + at , at : N ( 0,0.024
z156 = 0.49
z157 , z158 , z159 %95
.
: :
z156 ( 1) = 0.56, z156 ( 2 ) = 0.62, z156 ( 3) = 0.68
V e156 ( 1) = 0.024, V e156 ( 2 ) = 0.041, V e156 ( 2 ) = 0.054
%95 z157 , z158 , z159 21

12
= 0.56 1.96 0.024 = 0.56 0.304

= 0.62 1.96 0.041 = 0.62 0.397
= 0.68 1.96 0.054 = 0.68 0.455
12
12
1 z156 ( 1) u0.025 V e156 ( 1)
}
}( 3)
12
zn ( l) u 2 V en ( l)
{
{ V e
2 z156 ( 2 ) u0.025 V e156 ( 2 )

156
3 z156 ( 3) u0.025
) z157 ( 0.256,0.864 0.95 ) z158 ( 0.223,1.017

). z159 ( 0.225,1.135
63
64

Designing and Building
: Statistical Forecasting System
.
Iterative )
( .
: (Model Identification (Specification
) (
.
:
: : Variance-stabilizing Transformation
Time Plot

.41
: : d
d
:
-1 .
-2 SACF . SPACF
-3 1 2.

) ( .
: d 0 1 . 2
: p : q

p q

(ARMA(p,q 38
:
) (d,0,1 (AR(1
) (d,0,2 (AR(2
) (p,d,0 (AR(p
) (d,1,0 (MA(1
) (d,2,0 (MA(2
ACF

/
k = 0, k > 1
k = 0, k > 2
PACF
kk = 0, k > 1
kk = 0, k > 2
kk = 0, k > p
65
) (d,q ,0 (MA(q
) (d,1,1 (ARMA(1,1
) (p,d,q (ARMA(p,q
k = 0, k > q
/

1
1

q - p
q p
/ p q

/
p q
: :

w

12
s.e ( w ) 0 ( 1 + 2 r1 + 2 r2 + L + 2rK )
n
c0 = 0 r1 ,L, rK . K
H0 : = 0
H1 : 0
w
H 0 = 0.05 > 1.96
) s.e ( w
: Model Estimation
1 ,K , p 1 ,K , q
.
z1 , z2 ,K , zn 1, zn
2
) p ( B ) wt = + q ( B ) at , at : N ( 0, 2
p ( B ) zt = + q ( B ) at , at : N ( 0,
) p ( B ) q ( B p ( B ) = 0
) (.

.
: : The Method of Moments
r
z k
k
.
(AR(p:
n
-1 z = z = i =1 zi n
-2 1 ,K , p :
66
k = 1 k 1 + 2 k 2 + L + p k p , k > 1
(AR(p zt k .
k = 1, 2,K , p
Yule-Walker:
1 = 1 + 2 1 + L + p p 1
2 = 1 1 + 2 + L + p p 2
M
p = 1 p 1 + 2 p 2 + L + p
k rk 1 ,K ,p:
:
rp 1 1

rp 2 2
M M

1
p
r2 L rp 2
r1 L rp 3
M M M
rp 3 L r1
r1
r1 1
r r
1
2= 1
M M
M

rp rp 1 rp2

r1
r
2
M
rp

rp1
rp 2
M
r2 L rp 2
r1 L rp 3
M M M
rp 3 L r1
1 1
r1

1
2 r1
=
M M
M

rp 1 rp 2
p
2 = 0 1 1r1 2 r2 Lp rp
1
2
) ( zt z
n t =1
.
:
-1 (AR(1
= 0
) zt = 1 ( zt 1 ) + at , at : N ( 0, 2
1
1 = r1
= z
2 = 0 1 1r1
1 n
2
) 0 = ( zt z
n t =1
-2 (MA(1
67
) zt = at 1at 1 , at : N ( 0, 2
1
1 + 12
= 1

1
1 + 12
= r1
1
1 1 4 r1
= 1
2 r1
1 . 1 < 1 r1 = 0.4
( 1 ) 1 = 0.77 ( 1 ) 2 = 3.27 1 . 1 = 0.77
-3 (AR(2
) zt = 1 ( zt 1 ) + 2 ( zt 2 ) + at , at : N ( 0, 2
1 2
1 1 r1 r1
=
r1 1 r2
2
1

r rr
r r
1 = 1 122 , 2 = 2
1 r1
1 r
2
1
2
1
= z
2 = 0 1 1r1 2 r2
1
2
) ( zt z
n t =1
-4 (MA(2
= 0
zt = at 1at 1 2at 2 , at : N ( 0,
1 2
) 1 ( 1 2
2
= , 2
2
2
1 + 1 + 2
1 + 12 + 22
r1 r2 1 2
1 1 2
2
= r1
,
r
=
2
1 + 2 + 2
1 + 2 + 2
= 1
1 2 . 2 1 < 1, 2 + 1 < 1, 2 < 1

68
-5 (ARMA(1,1
zt = 1 ( zt 1 ) + at 1at 1 , at : N ( 0,
1 1
( 1 11 ) ( 1 1 ) , = ( 1 11 ) ( 1 1 )
2
1
2
2
= 1
1 + 1 211
1 + 1 211
r1 r2 1 1
1 11 1 1
1 11 1 1
= r1
= , r2
1
2
2
1 + 1 211
1 + 1 211
r2 r1
r
1 = 2
r1
. 1 1
()
()
r1
r2 r2
1 r 1 r 1
1
1
r1 =
r
1 + 12 2 2 1
r1
1 . 1 < 1
:
ARIMA(1,1,1), ARIMA(2,1,0), ARIMA(0,1,2), ARIMA(1,2,0),
.(ARIMA(0,2,1), ARIMA(0,2,0
: .
: Conditional Least Square

: Method
(ARMA(p,q
p ( B ) ( zt ) = q ( B ) at , at : N ( 0,
) p ( B ) q ( B q ( B ) = 0
) ( . at:
)p ( B
) ( zt
) q ( B
= at
} = { 1 , 2 ,K , p } = { 1 , 2 ,K , q
) (z
)
L pB
L pB
(1 B B
(1 B B
2
= ) at ( , ,
} z = { z1 , z2 ,K , zn
69
) a ( , , z
2
t
= ) min Sc ( , ,
, ,
t = p +1
Normal Equations .
=0
=
=
=
=0
=0
=
=
=
=
=
=
n 2
) at ( , , z
t = p+1
n 2
) at ( , , z
t = p+1
n 2
) at ( , , z
t = p+1
=
=
=
=
=
=
=
=
=
) Sc ( , ,
) Sc ( , ,
) Sc ( , ,
a p = a p1 = L = a p +1q = 0
) . .( t = p + 1
2
Sc , ,
)n ( p + q + 1
= 2
:
-1 (AR(1
} z = { z1 , z2 ,K , zn
zt = 1 ( zt 1 ) + at , at : N ( 0,
) zt z = 1 ( zt 1 z ) + at , at : N ( 0, 2
at ( 1 ) = ( zt z ) 1 ( zt 1 z ) , t = 2,3,L , n

a ( 1 ) = ( zt z ) 1 ( zt 1 z ) , t = 2,3,L , n
2
2
t
t =2
t =2
Sc ( 1 ) = at2 ( 1 ) = ( zt z ) 1 ( zt 1 z )
1 1
1 = 1
2
Sc ( 1 ) = at2 ( 1 ) = ( zt z ) 1 ( zt 1 z )
t =2
t =2
= ) Sc ( 1
( zt z ) 1 ( zt 1 z )
1
1 t =2
n
= 2 ( zt 1 z ) ( zt z ) 1 ( zt 1 z )
t =2
70
= 2 ( zt 1 z ) ( zt z ) 1 ( zt 1 z ) = 0
t =2
1 =1
) Sc ( 1
1
n
( zt 1 z ) ( zt z ) 1 ( zt 1 z ) = 0
t =2
=0
t =2
t =2
) ( zt 1 z ) ( zt z ) 1 ( zt1 z
) z ) ( zt z
2
. 1
: . 1
-2 (MA(1
)z
t 1
( z
t 1
( z
t =2
= 1
t =2
) zt = at 1at 1 , at : N ( 0, 2
z
xt = zt z
) xt = at 1at 1 , at : N ( 0, 2
at = xt 1at 1
x1 , x2 ,K , xn a0 = 0
a1 = x1
a2 = x2 1a1
a3 = x3 1a2
M
an = xn 1an 1
Sc ( 1 ) = at2
t =1
1 1 ) Sc ( 1
) (1,1- -
) at = at ( 1 1 *
) * dat (
) * dat (
d 1
d1
) * at ( 1 ) at ( * ) + ( 1
at = xt 1at 1
71
) da0 ( 1
= 0
d1
) dat ( 1 ) 1dat 1 ( 1
=
) + at 1 ( 1
d1
d1
.
) * dat (
d1
) + (
at ( 1 ) at (
1
n
Sc ( 1 ) = at2
t =1
1 *

. *
. .

(MA(1 .

:
-1 Maximum Likelihood Method
-2 Unconditional Least Squares Method
-3 Nonlinear Estimation Methods
: Model Checking and Diagnostics

) (4
at
. 2
et = zt zt = at , t = 1, 2,..., n
.

:
-1
-2
-3
2
-4 )
2
) ( at : IIDN ( 0,

: :
72
H 0 : E ( at ) = 0
H 1 : E ( at ) 0
= u

) se ( e
= 0.05 E ( at ) = 0 ) u < 1.96
30 (
: :
Runs test
)
241
(.
: :

Autocorrelation test
SACF
.
H 0 : 1 = 0
H1 : 1 0
r1

) se ( r1
1 = 0 . u < 1.96
: :

:
-1 Goodness of Fit Test
- Kolmogorov-
. Smirnov Test
-2 . Normal Probability Plot
-3 - . Q-Q Plot
:
(1 - Ljung-Box Q statistc LBQ
:
H 0 : 1 = 2 = L = K = 0
:
2
K
r
) Q = n ( n + 2) k : 2 ( K m
k =1 n k
m .
= u = 0.05
73
(2 Automatic Information Criteria AIC

:
2
AIC ( m ) = n ln a + 2m
) min AIC ( m
m
m
: Examples and Case Studies

-1
)z(t
59.9315
61.0344
60.6643
60.5777
59.9086
62.2658
61.9013
54.5045
55.8451
59.3572
61.8956
59.7252
58.4651
61.2797
58.0143
56.6085
56.8697
57.2961
59.6266
58.1105
61.9152
59.0778
58.2759
59.5015
63.4124
60.7196
62.8556
59.3191
59.5353
59.6004
59.1455
60.9492
58.1690
56.7437
63.8189
59.3402
59.9246
57.9624
57.8803
54.2185
57.0642
57.3940
59.3236
61.1107
58.7336
62.1957
56.9972
61.8685
56.6666
60.7356
57.2318
64.6886
60.5820
58.3755
60.1325
58.0151
59.5333
55.7339
58.9585
61.1520
59.0087
60.5289
59.1567
60.3373
55.4219
62.3728
60.1458
57.5307
61.5614
60.0377
60.8256
59.0780
63.1777
56.0309
59.2298
56.1346
63.5049
62.4654
59.3054
58.4174
59.0903
58.8802
63.7261
61.8370
61.6337
58.7564
60.8942
56.5413
61.4310
59.2086
60.8605
61.4451
53.8560
59.4119
59.3488
59.3839
59.8597
58.3583
56.1494
61.7218
56.4828
63.9622
59.1721
60.9225
58.1483
59.4554
61.7122
61.6708
58.5870
60.2990
58.2273
63.6776
54.6083
62.3827
57.8763
60.3843
63.5907
58.1711
59.9346
58.0423
55.4010
59.0997
59.5097
59.2927
61.1168
58.9275
63.1547
57.9813
61.1856
61.8108
59.1609
65.1325
62.6899
57.9363
59.9443
54.7163
60.0538
54.5550
63.3933
59.9569
60.0855
60.6919
61.1852
60.6201
60.7227
58.6501
59.0970
60.3563
60.2513
61.2179
59.5257
61.4230
58.7108
61.4761
62.1789
61.7008
60.5918
59.4444
57.3334
62.1017
57.3292
63.1070
58.1895
61.9205
56.2599
62.1362
58.4221
60.8962
60.3030
60.7021
58.5790
57.1459
60.9815
60.2052
60.9013
60.1815
61.0640
58.0059
61.2223
61.9753
60.4833
63.4411
59.3478
56.7241
58.1281
61.7840
60.9021
60.7001
61.9462
61.9448
60.9805
57.5151
59.2145
58.5278
60.3550
59.4242
60.3319
61.5555
59.7755
59.4755
Time Plot MINITAB

:
;')MTB > TSPlot 'z(t
>SUBC
;Index
>SUBC
;TDisplay 11
>SUBC
;Symbol
>SUBC
;Connect
>SUBC
Title "An obseved Time Series".
74
An obseved Time Series
z(t)
65
60
55
Index
50
100
150
200

MTB > %ACF 'z(t)';
SUBC>
MAXLAG 20;
SUBC>
TITLE"SACF of observed Time Series".
Executing from file: H:\MTBWIN\MACROS\ACF.MAC
Autocorrelation
SACF of observed Time Series

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag Corr
1
2
3
4
5
6
7
0.51
0.20
-0.00
-0.05
-0.08
-0.18
-0.19
LBQ
7.19
2.32
-0.01
-0.59
-0.95
-2.05
-2.09
52.48
60.78
60.78
61.34
62.82
69.92
77.58
10
Lag Corr
8
9
10
11
12
13
14
-0.14
-0.14
-0.09
-0.07
-0.08
-0.02
0.03
15
LBQ
-1.50
-1.52
-0.90
-0.71
-0.79
-0.21
0.32
81.76
86.14
87.73
88.73
89.97
90.05
90.27
Lag Corr
15
16
17
18
19
20
0.07
0.13
0.17
0.20
0.12
0.06
20
T
LBQ
0.68 91.23
1.33 94.86
1.75 101.33
2.06 110.63
1.21 113.98
0.61 114.86

MTB > %PACF 'z(t)';
SUBC>
MAXLAG 20;
SUBC>
TITLE"SPACF of obseved Time Series".
Executing from file: H:\MTBWIN\MACROS\PACF.MAC
75
SPACF of obseved Time Series

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag PAC
1
2
3
4
5
6
7
0.51
-0.08
-0.10
0.00
-0.05
-0.16
-0.04
10
T
7.19
-1.07
-1.38
0.04
-0.73
-2.34
-0.50
Lag PAC
8
9
10
11
12
13
14
-0.01
-0.12
0.01
-0.04
-0.09
0.03
0.02
15
T
Lag PAC
-0.12
-1.63
0.16
-0.60
-1.34
0.39
0.32
15
16
17
18
19
20
-0.02
0.09
0.08
0.06
-0.03
0.04
20
T
-0.23
1.28
1.20
0.86
-0.45
0.52

( AR(1
MTB > Name c7 = 'RESI1'
MTB > ARIMA 1 0 0 'z(t)' 'RESI1';
SUBC>
Constant;
SUBC>
Forecast 5 c4 c5 c6;
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
SUBC> GNormalplot.
ARIMA Model
ARIMA model for z(t)
Estimates at each iteration
Iteration
SSE
Parameters
0
839.667
0.100
53.870
1
746.819
0.250
44.876
2
695.840
0.400
35.883
3
685.086
0.502
29.769
4
685.054
0.507
29.458
5
685.054
0.507
29.443
Relative change in each estimate less than
Final Estimates of Parameters
Type
Coef
StDev
AR
1
0.5073
0.0611
Constant
29.4429
0.1309
Mean
59.7571
0.2656
0.0010
T
8.30
224.98
Number of observations: 201

Residuals:
SS = 685.020 (backforecasts excluded)
MS =
3.442 DF = 199
Modified Box-Pierce (Ljung-Box) Chi-Square statistic
Lag
12
24
36
Chi-Square
10.8(DF=11)
27.6(DF=23)
35.9(DF=35)
76
48
45.0(DF=47)
Forecasts from period 201

Period
202
203
204
205
206
Forecast
59.7079
59.7322
59.7445
59.7507
59.7539
95 Percent Limits
Lower
Upper
56.0707
63.3451
55.6537
63.8106
55.5600
63.9290
55.5394
63.9620
55.5357
63.9721
Actual
:
-1
zt = 59.76 + 0.51( zt 1 59.76) + at , at : WN ( 0,3.44 )
: t -2
( )
1 = 0.51, s.e. 1 = 0.061, t = 8.3

= 59.76, s.e. ( ) = 0.66
( )
= 29.44, s.e. = 0.131, t = 224.98

2 = 3.44, with d . f . = 199
:
-1
MTB > ZTest 0.0 1.855 'RESI1';
SUBC>
Alternative 0;
SUBC> GHistogram;
SUBC> GDotplot;
SUBC> GBoxplot.
Z-Test
Test of mu = 0.000 vs mu not = 0.000
The assumed sigma = 1.85
Variable
RESI1
N
201
Mean
-0.002
StDev
1.851
SE Mean
0.131
Z
-0.01
P
0.99

-2
MTB > Runs 0 'RESI1'.
Runs Test
RESI1
K =
0.0000
The observed number of runs = 94

The expected number of runs = 101.0796
107 Observations above K
94 below
The test is significant at 0.3149
Cannot reject at alpha = 0.05

77
-3
ACF of Residuals for z(t)
(with 95% confidence limits for the autocorrelations)
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
Lag
-4
PACF of Residuals for z(t)
(with 95% confidence limits for the partial autocorrelations)
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
Lag

: -5
-
78
Histogram of the Residuals

(response is z(t))
Frequency
30
20
10
0
-5
Residual
: .
Normal Probability Plot -
Normal Probability Plot for RESI1
99
Mean:
-1.6E-03
StDev:
1.85070
95
90
Percent
80
70
60
50
40
30
20
10
5
-5.0
-2.5
0.0
2.5
5.0
Data
:
K-S Test -
MTB > %NormPlot 'RESI1';
SUBC>
Kstest;
SUBC>
Title "Normal Test for Residuals".
Executing from file: H:\MTBWIN\MACROS\NormPlot.MAC
79
Normal Test for Residuals
.999
.99
Probability
.95
.80
.50
.20
.05
.01
.001
-5
RESI1
Average: -0.0016272
StDev: 1.85070
N: 201
Kolmogorov-Smirnov Normality Test

D+: 0.045 D-: 0.060 D : 0.060
Approximate P-Value: 0.074
H 0 : Residuals : N ( 0,3.44 )
H1 : Residuals N ( 0,3.44 )
-
+
D = 0.045, D = 0.06, D = 0.06

= 0.05 0.074 P-Value
.
:
5
Period
202
203
204
205
206
Forecast
59.7079
59.7322
59.7445
59.7507
59.7539
95 Percent Limits
Lower
Upper
56.0707
63.3451
55.6537
63.8106
55.5600
63.9290
55.5394
63.9620
55.5357
63.9721
Actual

Plot C4*C8 C5*C8 C6*C8;
SUBC>
Connect;
SUBC>
Type 1;
SUBC>
Color 1;
SUBC>
Size 1;
SUBC>
Title "Forecast of 5 future value with 95% limits";
SUBC>
Overlay.
80
Forecast of 5 future value with 95% limits

64
63
62
61
59
C4
60
58
57
56
55
5
C8
.
Forecast of 5 future value with 95% limits
64
63
62
61
59
C9
60
58
57
56
55
200
180
190
C8
-2
494.948
503.170
496.806
504.340
496.372
503.506
496.062
501.521
507.886
506.345
507.735
497.295
497.642
510.287
496.029
496.208
498.429
496.227
491.923
499.260
500.356
494.033
497.393
495.431
496.285
507.673
503.415
499.783
496.181
497.133
507.382
502.233
505.884
496.665
500.074
492.286
508.489
504.965
500.664
507.072
485.991
500.921
503.852
498.380
498.060
498.440
496.678
493.371
506.329
498.598
516.373
499.217
495.000
492.352
495.423
505.577
501.819
501.175
504.666
509.407
81
488.539
506.040
501.605
497.785
502.891
491.981
493.161
505.581
501.064
497.883
500.744
493.866
495.868
494.885
494.814
511.026
489.348
498.229
502.545
503.107
496.830
507.020
495.355
504.712
504.072
495.737
503.580
501.700
507.582
501.928
496.650
501.649
502.969
506.459
494.559
500.508
498.319
498.304
501.527
494.833
504.129
507.594
498.294
496.932
504.394
)z(t
499.148
503.975
498.758
493.057
499.890
507.416
498.090
504.877
494.918
499.173
496.765
482.567
500.989
501.331
489.314
503.134
501.634
499.611
500.388
505.987
493.858
494.857
497.861
493.017
498.931
497.794
513.900
504.717
502.209
497.992
501.387
494.878
497.943
499.299
498.787
499.467
502.933
495.430
511.741
501.999
496.078
497.413
495.353
499.134
503.130
504.174
501.701
500.199
504.785
505.517
505.475
497.462
504.817
504.336
490.887
503.465
506.649
505.218
498.477
496.680
491.202
495.699
492.716
504.408
501.417
492.318
496.757
498.403
491.707
505.429
514.220
502.414
491.418
496.150
498.016
501.542
507.590
497.647
504.640
495.954
493.552
497.130
497.626
506.259
504.085
490.683
487.344
494.614
507.438
512.122
505.367
504.012
497.231
505.234
496.877
499.982
498.640
502.173
504.346
504.913
500.712
504.204
506.403
502.721
503.746
506.001
501.827
496.063
505.709
493.502
497.421
497.048
503.325
500.484
502.066
501.082
493.843
498.571
501.703
498.599
500.256
497.643
490.619
500.324
499.779
497.956
502.681
502.823
502.720
502.489
495.691
496.252
499.279
498.169
497.015
503.195
500.252
498.158
500.409
496.225
491.726
489.032
500.024
505.194
499.574
: 50
510
)z(t
500
490
40
50
20
30
Index
10
: :
)Autocorrelation Function for z(t
15
20
LBQ
10
Lag Corr
0.05 0.60 113.33

-0.02 -0.26 113.47
0.07 0.82 114.85
-0.10 -1.16 117.65
0.07 0.75 118.81
-0.06 -0.73 119.93
: :
82
15
16
17
18
19
20
LBQ
5
Lag Corr
0.10 1.19 87.98

-0.11 -1.31 91.01
0.15 1.77 96.64
-0.18 -2.16 105.32
0.11 1.25 108.35
0.05 0.56 108.97
-0.12 -1.36 112.61
8
9
10
11
12
13
14
LBQ
Lag Corr
71.03
71.68
75.39
75.83
80.08
83.33
85.52
-8.38
-0.64
1.52
0.52
-1.61
1.39
-1.13
1 -0.53
2 -0.05
3 0.12
4 0.04
5 -0.13
6 0.11
7 -0.09
Autocorrelation
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Partial Autocorrelation Function for z(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag PAC
1
2
3
4
5
6
7
-0.53
-0.46
-0.29
-0.07
-0.11
0.02
-0.09
10
15
Lag PAC
-8.38
-7.28
-4.52
-1.08
-1.71
0.30
-1.36
8 0.04
9 -0.07
10 0.12
11 -0.07
12 -0.04
13 0.08
14 -0.04
0.65
-1.09
1.94
-1.15
-0.71
1.19
-0.59
Lag PAC
15
16
17
18
19
20
20
T
0.03 0.49
-0.13 -2.09
0.06 1.01
-0.08 -1.32
0.03 0.44
-0.13 -2.07
( MA(1
SUBC>
Constant;
SUBC>
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
SUBC> GNormalplot.
ARIMA Model
Iteration
SSE
Parameters
0
6081.19
0.100 500.046
1
5265.34
0.250 500.004
2
4615.22
0.400 499.980
3
4109.70
0.550 499.967
4
3766.60
0.700 499.960
5
3727.32
0.841 499.959
6
3687.70
0.797 499.963
7
3687.08
0.790 499.962
8
3687.07
0.791 499.962
9
3687.07
0.790 499.962
Type
Coef
StDev
MA
1
0.7905
0.0386
Constant
499.962
0.051
Mean
499.962
0.051
0.0010
T
20.50
9708.40

Residuals:
MS =
14.86 DF = 248
Lag
12
24
36
Chi-Square
26.7(DF=11)
35.9(DF=23)
63.1(DF=35)
83
48
82.8(DF=47)

Period
251
252
253
254
255
Forecast
502.256
499.962
499.962
499.962
499.962
95 Percent Limits
Lower
Upper
494.700
509.812
490.330
509.593
490.330
509.593
490.330
509.593
490.330
509.593
Actual
:
-1
zt = 499.962 + at 0.7905at 1 , at : WN ( 0,14.86 )
: t -2
( )
1 = 0.7905, s.e. 1 = 0.0386, t = 20.50
( )
= = 499.962, s.e. = 0.051, t = 9708.40

2 = 14.86, with d . f . = 248
:
1-
SUBC>
Alternative 0.
Z-Test
Variable
RESI1
N
250
Mean
-0.007
StDev
3.847
SE Mean
0.243
Z
-0.03
P
0.98

-2
Runs Test
RESI1
K =
0.0000

126 Observations above K 124 below

-3
84

1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
55
60
Lag
-4
1.0
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
55
60
Lag

: -5
-
(response is z(t))
30
Frequency
0.8
20
10
0
-10
10
Residual
: .
85
99
Mean:
-6.9E-03
StDev:
3.84651
95
90
Percent
80
70
60
50
40
30
20
10
5
-10
-5
10
Data
:
K-S Test -
MTB > %Qqplot 'RESI1';
SUBC>
Conf 95;
SUBC>
Ci.
Executing from file: H:\MTBWIN\MACROS\Qqplot.MAC
Distribution Function Analysis

Normal Dist. Parameter Estimates
Data
: RESI1
Mean:
StDev:
-6.9E-03
3.84651

SUBC>
Kstest.
Executing from file: H:\MTBWIN\MACROS\NormPlot.MAC
Normal Probability Plot
.999
.99
Probability
.95
.80
.50
.20
.05
.01
.001
-10
10
RESI1
Average: -0.0069004
StDev: 3.84651
N: 250

D+: 0.034 D-: 0.051 D : 0.051
-
86
D + = 0.034, D = 0.051, D = 0.051

P-Value 0.105 = 0.05
.
:
5
Actual
95 Percent Limits
Lower
Upper
494.700
509.812
490.330
509.593
490.330
509.593
490.330
509.593
490.330
509.593
Period
251
252
253
254
255
Forecast
502.256
499.962
499.962
499.962
499.962
95%
Forecast of 5 future values with 95% limits
510
C4
500
490
4
C8
-3
224.678
227.500
221.632
234.178
233.005
232.432
228.189
231.718
222.468
231.163
226.479
234.207
230.455
233.640
233.725
233.089
228.655
226.760
224.063
225.663
238.463
232.209
230.102
231.121
229.848
223.115
228.418
224.483
232.738
228.163
232.232
236.562
233.427
225.484
226.641
222.515
229.136
239.577
227.621
229.564
231.633
227.982
227.859
228.618
228.989
229.792
225.447
234.825
235.224
233.044
224.207
226.778
221.562
232.308
232.653
222.156
229.331
231.319
225.734
230.122
228.225
235.122
228.198
224.928
234.707
231.665
229.575
224.747
87
229.903
222.482
236.488
223.408
218.067
229.359
234.668
225.721
230.888
229.851
235.024
226.724
227.812
233.841
228.922
227.089
228.954
230.260
223.390
236.033
217.433
217.123
235.744
235.767
224.927
228.472
228.227
236.659
225.196
231.266
233.891
230.327
227.032
232.820
227.346
225.772
234.323
213.619
221.484
236.419
233.918
226.765
228.200
225.799
236.399
222.523
232.976
232.067
230.148
227.077
233.256
)z(t
229.574
224.077
230.713
215.405
228.758
234.678
234.155
227.075
230.421
224.663
233.335
223.571
234.561
232.473
231.653
230.146
233.444
220.958
231.531
231.363
228.285
229.320
235.587
228.729
232.700
240.884
236.017
229.321
228.141
228.958
226.114
220.782
232.030
230.938
229.757
230.636
230.713
230.360
234.625
241.821
230.865
226.222
225.985
229.352
228.837
222.819
232.315
233.127
229.733
227.948
227.046
230.662
234.107
238.762
230.499
225.771
225.358
229.369
232.881
226.865
232.027
233.852
232.160
224.258
228.606
234.243
231.265
233.628
226.994
225.616
225.291
230.296
236.029
232.152
232.135
232.684
235.038
223.994
228.203
239.883
229.486
230.342
223.795
226.271
223.250
228.920
235.339
232.461
228.727
231.411
233.799
225.273
231.898
243.963
225.070
227.643
223.381
230.088
225.447
227.934
235.210
230.076
221.171
230.582
232.344
224.880
227.449
239.660
225.860
229.792
235.112
238.292
227.805
230.794
231.092
225.020
231.062
224.050
232.032
233.622
223.599
229.100
242.860
225.123
230.082
228.468
235.623
226.745
229.727
232.891
224.096
Time Plot MINITAB

50) : (
242
)z(t
232
222
40
50
20
30
Index
10

15
20
LBQ
-1.65 376.32
-1.79 389.93
-1.42 398.81
-0.76 401.37
0.02 401.37
0.61 403.05
10
Lag Corr
-0.21
-0.22
-0.18
-0.10
0.00
0.08
15
16
17
18
19
20
LBQ
5
Corr
Lag
8 -0.02 -0.17 322.05

9 0.14 1.18 327.24
10 0.23 1.95 341.48
11 0.23 1.88 355.16
12 0.15 1.22 361.17
13 0.02 0.15 361.25
14 -0.12 -0.96 365.04
88
Lag Corr
LBQ
178.21
244.29
249.71
256.21
283.75
311.23
321.95
0.84 13.27
0.51 5.20
0.15 1.35
-0.16 -1.46
-0.33 -2.98
-0.33 -2.87
-0.20 -1.73
1
2
3
4
5
6
7
Autocorrelation
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag PAC
1
2
3
4
5
6
7
10
T
0.84 13.27
-0.66 -10.39
-0.07 -1.13
-0.06 -1.00
0.11
1.78
0.14
2.18
-0.01 -0.22
Lag PAC
8
9
10
11
12
13
14
0.07
-0.07
0.05
-0.07
0.06
-0.15
0.01
15
T
1.06
-1.12
0.73
-1.18
0.90
-2.41
0.18
Lag PAC
15
16
17
18
19
20
0.04
-0.06
0.04
-0.07
0.09
-0.09
20
T
0.67
-0.93
0.60
-1.14
1.44
-1.39

( AR(2
SUBC>
Constant;
SUBC>
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
SUBC> GNormalplot.
ARIMA Model
Iteration
SSE
Parameters
0
4257.23
0.100
0.100 183.784
1
3528.31
0.250
0.012 169.535
2
2889.23
0.400
-0.076 155.360
3
2338.97
0.550
-0.165 141.201
4
1877.39
0.700
-0.253 127.051
5
1504.46
0.850
-0.342 112.913
6
1220.13
1.000
-0.430
98.789
7
1024.34
1.150
-0.519
84.690
8
916.97
1.300
-0.608
70.623
9
894.38
1.402
-0.668
61.154
10
894.31
1.408
-0.672
60.670
11
894.31
1.408
-0.672
60.646
Relative change in each estimate less than 0.0010
Type
Coef
StDev
AR
1
1.4079
0.0473
AR
2
-0.6720
0.0474
Constant
60.6458
0.1203
Mean
229.638
0.456
T
29.78
-14.19
504.11

Residuals:
MS =
3.618 DF = 247
89

Lag
12
24
36
Chi-Square
17.5(DF=10)
27.2(DF=22)
49.7(DF=34)
48
67.7(DF=46)

Period
251
252
253
254
255
256
257
258
259
260
Forecast
224.939
226.747
228.725
230.296
231.177
231.363
231.033
230.442
229.833
229.372
95 Percent Limits
Lower
Upper
221.211
228.668
220.308
233.186
220.642
236.808
221.546
239.045
222.311
240.044
222.494
240.233
222.070
239.996
221.327
239.558
220.600
239.067
220.090
238.655
Actual
:
-1
zt = 60.6458 + 1.4079 zt 1 0.672 zt 2 + at , at : WN ( 0,3.618 )
( )
s.e. ( ) = 0.0474,
: t -2
1 = 1.4079, s.e. 1 = 0.0473, t = 29.78

2 = 0.672,
t = 14.19
= 229.638, s.e. ( ) = 0.456
( )
= 60.6458, s.e. = 0.1203, t = 504.11

2 = 3.618, with d . f . = 247
:
-1
SUBC>
Alternative 0.
Z-Test
Variable
RESI1
N
250
Mean
-0.005
StDev
1.894
SE Mean
0.229
Z
-0.02
P
0.98

-2
Runs Test
RESI1
K =
0.0000
90

129 Observations above K 121 below

-3
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
55
60
Lag
-4
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
50
55
60
Lag

: -5
-
91

(response is z(t))
30
Frequency
20
10
0
-5
Residual
: .
99
Mean:
-4.6E-03
StDev:
1.89436
95
90
Percent
80
70
60
50
40
30
20
10
5
-6
-4
-2
Data
:
K-S Test -
.999
.99
Probability
.95
.80
.50
.20
.05
.01
.001
-5
RESI1
Average: -0.0046305
StDev: 1.89436
N: 250

D+: 0.020 D-: 0.029 D : 0.029
Approximate P-Value > 0.15
0.15 P-Value -
. = 0.05
92
:
10
Period
251
252
253
254
255
256
257
258
259
260
95 Percent Limits
Lower
Upper
221.211
228.668
220.308
233.186
220.642
236.808
221.546
239.045
222.311
240.044
222.494
240.233
222.070
239.996
221.327
239.558
220.600
239.067
220.090
238.655
Forecast
224.939
226.747
228.725
230.296
231.177
231.363
231.033
230.442
229.833
229.372
Actual
95%
C4
240
230
220
0
10
C8

245
C9
235
225
215
0
100
200
C8
93
240
C9
230
220
60
40
50
20
30
10
C8

.
:
)
(
2.09
1.57
2.07
1.78
2.00
1.42
1.82
1.68
2.83
1.46
2.08
1.37
1.76
1.54
1.85
1.79
3.44
2.05
1.42
1.15
2.40
2.50
1.39
1.25
1.95
2.25
1.18
1.61
2.70
1.58
1.27
1.77
1.54
1.25
1.08
2.91
Defects
1.20 1.50
1.89 1.80
1.40 1.51
2.32 1.23
1.84
3.5
3.0
2.0
Defects
2.5
1.5
1.0
40
20
30
94
10
Index
Autocorrelation
Autocorrelation Function for Defects

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag
Corr
LBQ
Lag
Corr
LBQ
1 0.43 2.88
2 0.26 1.49
3 0.14 0.77
4 0.08 0.43
5 -0.09 -0.46
6 -0.07 -0.39
7 -0.21 -1.10
8.84
12.18
13.18
13.50
13.89
14.18
16.57
8
9
10
11
-0.11
-0.05
-0.01
-0.04
-0.57
-0.27
-0.04
-0.19
17.25
17.41
17.41
17.50
10
11
Partial Autocorrelation Function for Defects

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag PAC
1
2
3
4
5
6
7
5
T
0.43 2.88
0.09 0.63
-0.00 -0.01
0.00 0.00
-0.16 -1.07
0.00 0.02
-0.18 -1.19
Lag PAC
8
9
10
11
10
11
0.07 0.44
0.05 0.35
0.01 0.09
-0.03 -0.23
AIC
:
2
AIC ( m ) = n ln a + 2m
min AIC ( m )
m
m
:
MTB > ARIMA 1 0 0 'Defects' 'RESI1';
SUBC>
Constant;
ARIMA Model
ARIMA model for Defects
Type
Coef
StDev
AR
1
0.4421
0.1365
T
3.24
95
Constant
Mean
0.99280
1.7795
0.06999
0.1254
14.19
Residuals:
MS = 0.22042 DF = 43
Lag
12
24
36
Chi-Square
4.9(DF=11)
8.9(DF=23)
30.9(DF=35)
48
* (DF= *)

SUBC>
Constant;
ARIMA Model
Type
Coef
StDev
AR
1
0.3999
0.1533
AR
2
0.0989
0.1531
Constant
0.89019
0.07047
Mean
1.7762
0.1406
T
2.61
0.65
12.63
Residuals:
MS = 0.22347 DF = 42
Lag
12
24
36
Chi-Square
4.0(DF=10)
8.1(DF=22)
28.8(DF=34)
48
* (DF= *)

SUBC>
Constant;
ARIMA Model

Type
Coef
StDev
AR
1
0.5983
0.2691
MA
1
0.1926
0.3294
Constant
0.71334
0.05693
Mean
1.7759
0.1417
T
2.22
0.58
12.53
Residuals:
MS = 0.22367 DF = 42
Lag
12
24
36
Chi-Square
4.1(DF=10)
8.3(DF=22)
29.1(DF=34)
SUBC>
Constant;
96
48
* (DF= *)
ARIMA Model

Type
Coef
StDev
MA
1
-0.3409
0.1431
Constant
1.78480
0.09651
Mean
1.78480
0.09651
T
-2.38
18.49
Residuals:
MS = 0.2334 DF = 43
Lag
12
24
36
Chi-Square
8.0(DF=11)
13.2(DF=23)
35.7(DF=35)
48
* (DF= *)

SUBC>
Constant;
ARIMA Model
Type
Coef
StDev
MA
1
-0.3869
0.1516
MA
2
-0.1816
0.1516
Constant
1.7839
0.1118
Mean
1.7839
0.1118
T
-2.55
-1.20
15.96
Residuals:
MS = 0.22882 DF = 42
Lag
12
24
36
Chi-Square
4.6(DF=10)
9.2(DF=22)
31.0(DF=34)
SUBC>
Constant;
ARIMA Model
Type
Coef
StDev
AR
1
0.4134
1.5680
AR
2
0.0929
0.7113
MA
1
0.0136
1.5749
Constant
0.87675
0.07036
Mean
1.7761
0.1425
T
0.26
0.13
0.01
12.46
97
48
* (DF= *)
Residuals:
MS = 0.22892 DF = 41
Lag
12
24
36
Chi-Square
4.0(DF= 9)
8.1(DF=21)
28.8(DF=33)
48
* (DF= *)

SUBC>
Constant;
ARIMA Model
* ERROR * Model cannot be estimated with these data
SUBC>
Constant;
ARIMA Model
Type
Coef
StDev
AR
1
1.6720
0.1165
AR
2
-0.7263
0.1251
MA
1
1.3199
0.0184
MA
2
-0.3196
0.0731
Constant 0.096224
0.003323
Mean
1.77238
0.06121
T
14.35
-5.80
71.63
-4.37
28.95
Residuals:
MS = 0.20831 DF = 40
Lag
12
24
36
Chi-Square
4.7(DF= 8)
8.9(DF=20)
29.7(DF=32)
48
* (DF= *)
98
2
m
AIC
_________ ___ ________
0.22042
3
62.0499
0.22347
4
59.4315
0.23340
3
59.4751
0.22882
4
58.3669
0.22367
4
59.3913
0.22892
5
56.3472
0.20831
6
58.5928
Model
__________
)AR ( 1
) AR ( 2
)MA ( 1
) MA ( 2
)ARMA ( 1,1
)ARMA ( 2,1
) ARMA ( 1, 2
) ARMA ( 2,2
min AIC ( m ) = 62.0499

m
. (AR(1
.
:

5.43
3.88
3.57
2.75
6.06
5.80
5.16
6.64
6.43
7.86
3.80
4.30
3.45
4.80
6.12
6.08
5.71
7.49
7.53
7.50
4.60
3.91
3.98
5.43
5.52
5.67
5.63
6.64
7.59
8.75
4.14
5.42
1.22
3.08
5.65
4.78
5.61
6.09
5.62
8.27
4.77
5.07
2.65
4.40
4.79
4.89
5.70
4.72
7.27
8.50
3.45
3.78
3.28
3.84
6.11
4.99
5.75
6.08
9.01
7.23
3.99
6.16
4.05
5.14
6.46
6.12
6.36
6.57
7.49
7.53
5.51
4.05
4.08
4.00
4.31
6.23
8.02
7.56
6.69
7.67
5.74
2.54
4.61
1.58
4.99
5.05
7.07
6.87
7.22
8.22
9
8
7
Sales
4
3
2
1
100
90
80
70
60
50
99
40
30
20
10
Index
Sales
3.49
3.96
2.89
2.52
5.77
3.20
5.13
7.20
7.26
6.42

Autocorrelation Function for Sales
15
25
LBQ
Lag Corr
0.67 565.04
0.50 568.77
0.64 574.90
0.75 583.43
0.22
0.17
0.21
0.25
22
23
24
25
LBQ
Autocorrelation
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag Corr
1.32 483.85
1.13 499.04
0.97 510.68
0.92 521.52
1.11 537.81
0.95 550.11
0.77 558.52
0.41
0.35
0.31
0.30
0.36
0.31
0.26
15
16
17
18
19
20
21
LBQ
Lag Corr
2.22 317.32
1.87 344.49
1.79 371.67
1.82 401.71
1.46 422.60
1.29 439.91
1.50 464.06
0.56
0.49
0.49
0.51
0.42
0.38
0.45
8
9
10
11
12
13
14
LBQ
7.10 51.97
4.26 89.85
3.94 134.64
3.36 177.75
2.83 215.72
2.63 253.97
2.12 282.61
Lag Corr
0.71
0.60
0.65
0.64
0.59
0.59
0.51
1
2
3
4
5
6
7
Partial Autocorrelation Function for Sales
25
15
T
Lag PAC
22 -0.17 -1.68
23 -0.10 -1.00
24 0.15 1.55
25 0.03 0.34
5
Lag PAC
0.32
-0.96
-0.84
-0.10
1.43
-0.44
0.04
0.03
-0.10
-0.08
-0.01
0.14
-0.04
0.00
15
16
17
18
19
20
21
Lag PAC
8 0.19 1.93
9 -0.17 -1.70
10 0.14 1.44
11 0.03 0.34
12 -0.15 -1.52
13 0.02 0.25
14 0.04 0.39
Lag PAC
1 0.71 7.10
2 0.20 1.99
3 0.35 3.53
4 0.15 1.49
5 0.09 0.92
6 0.10 1.03
7 -0.13 -1.27
.
wt = zt
3
2
1
)w(t
0
-1
-2
-3
100
90
80
70
60
50
40
30
20
10
Index
.
100
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Autocorrelation
Autocorrelation Function for w(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
Lag Corr
12
LBQ
1 -0.30 -3.00
2 -0.31 -2.86
3 0.11 0.90
4 0.04 0.31
5 -0.04 -0.33
6 0.14 1.15
7 -0.24 -1.98
9.26
19.32
20.49
20.63
20.80
22.80
29.02
Lag Corr
LBQ
8 0.19 1.53
9 -0.07 -0.53
10 -0.07 -0.52
11 0.16 1.27
12 -0.07 -0.53
13 -0.14 -1.04
14 0.15 1.10
33.07
33.58
34.08
37.12
37.67
39.85
42.37
22
Lag Corr
LBQ
15 0.00 0.00
16 0.02 0.13
17 -0.07 -0.52
18 -0.15 -1.09
19 0.20 1.48
20 0.01 0.05
21 -0.04 -0.25
42.37
42.40
42.99
45.65
50.73
50.74
50.90
Lag Corr
LBQ
22 0.05 0.35 51.22

23 -0.18 -1.28 55.44
24 0.03 0.20 55.55
Partial Autocorrelation Function for w(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
12
Lag PAC
Lag PAC
1 -0.30
2 -0.44
3 -0.22
4 -0.21
5 -0.17
6 0.06
7 -0.26
-3.00
-4.41
-2.23
-2.05
-1.72
0.59
-2.55
8 0.12
9 -0.16
10 -0.06
11 0.09
12 -0.06
13 -0.02
14 -0.07
1.17
-1.59
-0.60
0.93
-0.59
-0.17
-0.71
22
Lag PAC
15
16
17
18
19
20
21
0.06
0.07
-0.03
-0.17
-0.03
-0.10
0.09
T
0.59
0.70
-0.32
-1.65
-0.31
-0.95
0.85
Lag PAC
22 0.08 0.80
23 -0.15 -1.53
24 -0.05 -0.55
AIC
:
2
AIC ( m ) = n ln a + 2m
min AIC ( m )
m
m
:
MTB > ARIMA 1 1 0 'Sales';
SUBC>
NoConstant.
ARIMA Model
ARIMA model for Sales
Type
Coef
StDev
AR
1
-0.3114
0.0959
T
-3.25
101
Differencing: 1 regular difference

Number of observations: Original series 100, after differencing 99
Residuals:
MS =
1.359 DF = 98
Lag
12
24
36
Chi-Square
31.9(DF=11)
51.2(DF=23)
62.8(DF=35)
48
81.0(DF=47)

SUBC>
NoConstant.
ARIMA Model
Type
Coef
StDev
AR
1
-0.4532
0.0897
AR
2
-0.4656
0.0901
T
-5.05
-5.17

Residuals:
MS =
1.080 DF = 97
Lag
12
24
36
Chi-Square
21.8(DF=10)
40.9(DF=22)
49.4(DF=34)
48
59.9(DF=46)

SUBC>
NoConstant.
ARIMA Model
Type
Coef
StDev
MA
1
0.7636
0.0648
T
11.78

Residuals:
MS =
1.035 DF = 98
Lag
12
24
36
Chi-Square
12.6(DF=11)
27.8(DF=23)
35.9(DF=35)
SUBC>
NoConstant.
ARIMA Model
102
48
48.5(DF=47)

Type
Coef
StDev
MA
1
0.5756
0.0990
MA
2
0.2029
0.0998
T
5.81
2.03

Residuals:
MS = 1.0232 DF = 97
Lag
12
24
36
Chi-Square
14.3(DF=10)
28.3(DF=22)
36.5(DF=34)
48
47.0(DF=46)

SUBC>
NoConstant.
ARIMA Model
Type
Coef
StDev
AR
1
0.1283
0.1334
MA
1
0.8027
0.0799
T
0.96
10.04

Residuals:
MS =
1.035 DF = 97
Lag
12
24
36
Chi-Square
13.3(DF=10)
27.9(DF=22)
36.1(DF=34)
48
48.2(DF=46)

SUBC>
NoConstant.
ARIMA Model
* WARNING * Back forecasts not dying out rapidly

Type
Coef
StDev
AR
1
-1.1389
0.0987
AR
2
-0.1440
0.0983
MA
1
-0.9889
0.0002
T
-11.53
-1.47
-3987.49

Residuals:
MS =
1.398 DF = 96
Lag
12
24
36
103
48
Chi-Square
35.1(DF= 9)
53.5(DF=21)
66.6(DF=33)
83.2(DF=45)

SUBC>
NoConstant.
ARIMA Model
Type
Coef
StDev
AR
1
-0.3476
0.4077
MA
1
0.2422
0.3771
MA
2
0.4506
0.2656
T
-0.85
0.64
1.70

Residuals:
MS = 1.0129 DF = 96
Lag
12
24
36
Chi-Square
11.9(DF= 9)
25.0(DF=21)
32.1(DF=33)
48
41.8(DF=45)

SUBC>
NoConstant.
ARIMA Model
Type
Coef
StDev
AR
1
-0.0691
0.3618
AR
2
-0.2941
0.1450
MA
1
0.5637
0.3737
MA
2
0.0840
0.3266
T
-0.19
-2.03
1.51
0.26

Residuals:
MS = 0.9857 DF = 95
Lag
12
24
36
Chi-Square
11.0(DF= 8)
23.4(DF=20)
30.1(DF=32)
48
36.5(DF=44)
104
2
m
AIC
_________ ___ ________
1.359
2
34.368
1.080
3
13.619
1.035
2
7.4057
1.023
3
8.2706
1.035
3
9.4057
1.398
4
41.169
1.013
4
9.2689
0.986
5
8.5741
Model
__________
)ARI ( 1,1
)ARI ( 2,1
)IMA ( 1,1
) IMA ( 1,2
)ARIMA ( 1,1,1
)ARIMA ( 2,1,1
) ARIMA ( 1,1, 2
) ARIMA ( 2,1, 2
min AIC ( m ) = 7.406

m
. (IMA(1,1 .
105
Seasonal Autoregressive Integrated Moving

Average Models

.

Seasonal Time Series
70
)z(t
60
50
150
100
Index
50
Seasonal Time Series

1000
900
)z(t
800
700
600
150
100
Index
50

SARIMA(p,d,q)(P,D,Q)s
SARIMA(0,1,1)(1,1,0)12
) at : WN ( 0, 2
106
(1 B ) (1 B) z = (1 B) a ,
s
) p,d,q)(P,D,Q)s
SARIMA(p,d,q)(P,D,Q)s
) p ( B ) P ( B s ) ( 1 B ) ( 1 B s ) zt = + q ( B ) Q ( B s ) at , at : WN ( 0, 2
D
) p ( B ) q ( B
s
s
2s
Ps
P ( B ) = 1 + 1B + 2 B + L + P B
Q ( B ) = 1 + 1B + 2 B + L + Q B
.Multiplicative Seasonal Models
: ) at : WN ( 0, 2
Qs
2s
:
D
d
wt = ( 1 B ) ( 1 B s ) zt
SARMA(p,q)(P,Q)s
) p ( B ) P ( B ) wt = + q ( B ) Q ( B ) at , at : WN ( 0, 2
s
SARMA(0,1)(1,1)12
) wt = wt 12 + at at 1 at 12 + at 13 , at : WN ( 0, 2
wt
0 = 12 + + ( ) 2 + ( + ) 2
2
= 12 + 2 ( 1 + 2 ) + ( ) ( 1 + 2 )
= 12 + 2 ( 1 + 2 ) 1 + ( )
wt 12
12 = 0 2 + ( ) 2
) = 0 2 ( 1 + 2

1 + 2
1 2
2
( )
2
2
12 = ( 1 + ) +
1 2
) 0 = 2 (1+ 2
) 1 = E ( wt wt 1
) = 11 2 E ( at 12 wt 1 ) + E ( at 13wt 1
= 11 2 + ( ) 2
107
11 = E ( wt wt 11 ) = 1 +
( ) 2
1 = 1 +
1 2
( )
11 = 2
1 2
2 = 3 = L = 10 = 0
13 = 11
k = k 12 , k > 13
k =0
1,

,
k =1
2
1+
0,
k = 2,...,10
( ) ( 1 )
k = k =
, k = 11
0 1 + 2 1 + 2 2
( ) ( 1 )
,
k = 12
2
1
+
k = 13
11 ,
,
k > 13
k 12
wt = ( 1 B s ) at SARIMA(0,d,0)(0,D,1)s -1
k =0
1,
k =
,
k=s
2
1
+
otherwise
0,
( 1 B ) w
s
= at
SARIMA(0,d,0)(1,D,1)s -2
k =0
1,
ks
k = , k = s, 2 s,...
0,
otherwise
wt = ( 1 B ) ( 1 B s ) at
108
SARIMA(0,d,1)(0,D,1)s -3
k =0
1,
,
k =1
2
1+
, k = s 1
2
2
k = ( 1 + ) ( 1 + )
,
k=s
2
1+
s 1 ,
k = s +1
otherwise
0,
( 1 B ) w = ( 1 B ) a
s
k =0
1,
( ) ( 1 ) k s 1
k =
, k = s, 2 s,...
2
1 + 2
otherwise
0,
( 1 B s ) wt = ( 1 B ) at
SARIMA(0,d,0)(1,D,1)s -4
SARIMA(0,d,1)(1,D,0)s -5
k =0
1,
, k =1
2
1+
0,
k = 2,..., s 2
k =
, k = s 1
1+ 2
k=s
,
s 1 ,
k = s +1
k > s +1
k s ,
wt = ( 1 1B 2 B 2 ) ( 1 B12 ) at
109
SARIMA(0,d,2)(0,D,1)s -6
1,
1 ( 1 2 ) ,
1 + 12 + 22
,
1 + 12 + 22
,
2
( 1 + 1 + 22 ) ( 1 + 2 )
k =
1 ( 1 2 )
,
( 1 + 2 + 2 ) ( 1 + 2 )
1
2
2
1
+
s 1 ,
,
s 2
0,
k =0
k =1
k =2
k = s2
k = s 1
k=s
k = s +1
k = s+2
otherwise

.
: SARIMA(0,d,1)(1,D,0)12
= 0.6, = 0.5
(1)
ACF of SARIMA(0,d,1)(1,D,0)12
C
1
0.5
0.0
-0.5
10
20
30
40
50
Lag
= 0.6, = 0.5 ( 2)
0.7
0.6
C
1
0.5
0.4
0.3
0.2
0.1
0.0
0
10
20
30
40
50
Lag
= 0.6, = 0.5
110
(3)
0.5
C
1
0.0
-0.5
50
40
20
30
10
Lag
)(4
= 0.6, = 0.5
0.5
C
1
0.0
-0.5
50
40
20
30
10
Lag
:

. cut off
:
-1 wt = ( 1 B12 ) at
( = 0.6
0.0
-0.1
-0.2
-0.4
-0.5
-0.6
50
40
20
30
Lag
= 0.6
111
10
C
1
-0.3
0.6
0.5
0.4
0.2
C
1
0.3
0.1
0.0
-0.1
-0.2
40
50
20
30
10
Lag
-2
( = 0.6
( 1 B ) w = a
12
0.6
0.5
0.4
C
1
0.3
0.2
0.1
0.0
40
50
20
30
10
Lag
( = 0.6
0.0
-0.1
-0.2
C
1
-0.3
-0.4
-0.5
-0.6
50
40
20
30
10
Lag
: )
(
54.9
58.2
68.2
54.9
53.2
57.3
70.4
53.4
51.7
54.9
54.3
70.1
57.5
52.8
52.7
69.9
55.3
51.5
55.3
53.4
67.5
61.2
52.8
51.6
61.0
58.2
52.3
56.9
53.0
58.0
69.3
54.5
52.4
52.7
66.9
53.6
57.4
52.8
54.2
69.8
56.4
52.1
53.9
70.7
55.3
61.5
53.3
53.2
66.1
59.3
52.6
53.3
65.3
58.5
72.7
54.4
53.0
56.1
68.7
53.9
53.5
56.5
69.3
72.2
56.0
53.0
53.6
70.0
56.4
53.4
53.4
69.6
112
71.5
60.0
53.4
53.0
67.9
61.7
53.6
52.5
64.2
59.1
71.0
54.6
52.8
58.7
68.3
55.1
53.2
55.5
57.2
70.6
55.6
52.6
55.4
67.9
57.3
53.0
53.3
56.3
68.2
59.4
52.9
52.9
65.3
61.9
53.5
52.4
55.8
57.7
69.8
54.0
53.0
58.2
69.9
54.3
51.5
55.7
54.6
71.0
54.9
52.7
55.8
71.5
56.3
51.5
)z(t
56.3
54.9
67.4
56.6
53.4
55.3
65.1
59.4
52.1
51.5 52.2 57.1 63.6 68.8 68.9 60.1 55.6 53.9 53.3 53.1 53.5 53.5 53.9 57.1
64.7 69.4 70.3 62.6 57.9 55.8 54.8 54.2 54.6 54.3 54.8 58.1 68.1 73.3 75.5
66.4 60.5 57.7 55.8 54.7 55.0 55.6 56.4 60.6 70.8 76.4 74.8 62.2
z(t)
70
60
50
Index
50
100
150

Autocorrelation
Autocorrelation Function for z(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
12
Lag
Corr
LBQ
Lag
Corr
1
2
3
4
5
6
7
8
9
10
11
12
0.77
0.33
-0.11
-0.38
-0.52
-0.56
-0.52
-0.38
-0.12
0.29
0.69
0.88
10.31
2.93
-0.99
-3.28
-4.22
-4.16
-3.50
-2.39
-0.71
1.76
4.15
4.86
108.10
127.34
129.74
156.83
207.24
266.55
316.80
343.71
346.25
362.08
453.58
603.87
13
14
15
16
17
18
19
20
21
22
23
24
0.68
0.28
-0.12
-0.36
-0.49
-0.53
-0.49
-0.37
-0.14
0.23
0.61
0.79
22
T
32
LBQ
Lag
Corr
3.34 694.28
1.30 709.82
-0.53 712.43
-1.65 738.30
-2.21 786.28
-2.33 843.04
-2.10 892.36
-1.54 920.39
-0.56 924.24
0.95 935.37
2.46 1011.31
3.10 1140.90
25
26
27
28
29
30
31
32
33
34
35
36
0.61
0.25
-0.11
-0.34
-0.47
-0.51
-0.48
-0.36
-0.14
0.19
0.54
0.72
42
LBQ
Lag
Corr
2.30 1220.07
0.91 1233.18
-0.42 1235.98
-1.24 1261.29
-1.67 1308.32
-1.80 1364.72
-1.65 1414.08
-1.23 1442.71
-0.49 1447.34
0.66 1455.79
1.82 1520.89
2.37 1636.49
37
38
39
40
41
42
43
44
0.57
0.23
-0.11
-0.33
-0.44
-0.48
-0.45
-0.34
LBQ
1.82 1709.36
0.73 1721.70
-0.34 1724.43
-1.02 1749.06
-1.38 1795.00
-1.48 1849.41
-1.37 1897.05
-1.04 1925.30


1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
12
22
32
42
Lag
PAC
Lag
PAC
Lag
PAC
Lag
PAC
1
2
3
4
5
6
7
8
9
10
11
12
0.77
-0.68
-0.06
0.01
-0.42
-0.18
-0.11
-0.22
0.18
0.57
0.26
0.16
10.31
-9.01
-0.82
0.07
-5.55
-2.45
-1.48
-2.96
2.39
7.65
3.50
2.16
13
14
15
16
17
18
19
20
21
22
23
24
-0.42
0.29
0.01
-0.07
0.07
-0.01
-0.09
-0.05
-0.08
0.10
-0.03
0.03
-5.62
3.89
0.16
-0.92
0.95
-0.11
-1.15
-0.72
-1.11
1.38
-0.35
0.37
25
26
27
28
29
30
31
32
33
34
35
36
-0.13
-0.00
0.04
-0.10
-0.03
0.01
0.01
-0.04
-0.04
0.01
-0.02
0.07
-1.75
-0.05
0.50
-1.30
-0.34
0.19
0.07
-0.55
-0.52
0.15
-0.32
0.90
37
38
39
40
41
42
43
44
-0.06
-0.05
0.01
0.00
-0.05
0.05
-0.05
-0.03
-0.74
-0.62
0.17
0.00
-0.63
0.66
-0.64
-0.42
.
113

z(t)
589
673
678
621
713
796
801
747
826
898
908
827
561
742
639
602
667
858
764
711
799
957
867
797
640
716
604
635
762
826
725
751
890
924
815
843
656
660
611
677
784
783
723
804
900
881
812
727
617
594
635
837
740
690
756
961
837
773
697
583
634
736
817
701
734
860
935
784
813
640
587
658
755
767
706
750
878
894
791
834
599
565
622
811
722
677
707
942
855
760
782
568
598
709
798
681
711
807
913
809
802
892
577
628
722
735
687
734
824
869
810
828
903
553
618
782
697
660
690
886
834
766
778
966
582
688
756
661
698
785
859
790
805
889
937
600
705
702
667
717
805
819
800
821
902
896
566
770
653
645
696
871
783
763
773
969
858
653
736
615
688
775
845
740
800
883
947
817
1000
z(t)
900
800
700
600
Index
50
100
150

Autocorrelation

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
Lag
1
2
3
4
5
6
7
8
9
10
11
12
Corr
12
LBQ
Lag
Corr
0.89 11.56
0.78 6.27
0.62 4.12
0.49 2.95
0.43 2.47
0.38 2.10
0.41 2.25
0.45 2.40
0.56 2.87
0.69 3.34
0.77 3.52
0.84 3.61
135.94
240.13
306.72
347.97
380.09
405.02
435.54
472.37
529.07
614.27
721.72
852.41
13
14
15
16
17
18
19
20
21
22
23
24
0.74
0.64
0.49
0.36
0.31
0.25
0.29
0.32
0.42
0.53
0.60
0.67
22
LBQ
Lag
Corr
2.96 954.68
2.411030.09
1.791074.85
1.311099.68
1.091117.39
0.901129.76
1.011145.59
1.121165.42
1.441199.13
1.811253.81
2.031325.51
2.211415.29
25
26
27
28
29
30
31
32
33
34
35
36
0.58
0.49
0.35
0.24
0.19
0.14
0.17
0.20
0.28
0.38
0.45
0.52
32
T
42
LBQ
Lag
Corr
1.861483.11
1.521530.99
1.091556.45
0.731568.20
0.571575.44
0.431579.55
0.511585.52
0.601593.62
0.851610.35
1.151641.57
1.351685.27
1.531742.95
37
38
39
40
41
42
0.43
0.35
0.22
0.12
0.06
0.02
LBQ
1.271784.17
1.001810.60
0.641821.68
0.331824.67
0.181825.56
0.051825.62
114

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
Lag
12
22
32
42
PAC
Lag
PAC
Lag
PAC
Lag
PAC
1 0.89
2 -0.08
3 -0.28
4 0.03
5 0.35
6 -0.08
7 0.28
8 0.09
9 0.40
10 0.30
11 0.06
12 0.22
11.56
-1.06
-3.65
0.42
4.54
-1.07
3.67
1.19
5.17
3.95
0.81
2.88
13
14
15
16
17
18
19
20
21
22
23
24
-0.63
-0.02
0.07
-0.04
-0.09
-0.04
-0.05
0.03
0.04
0.05
0.05
0.05
-8.19
-0.21
0.95
-0.52
-1.12
-0.49
-0.60
0.38
0.46
0.67
0.60
0.59
25
26
27
28
29
30
31
32
33
34
35
36
-0.18
0.08
0.06
-0.03
-0.04
0.00
-0.06
-0.01
-0.01
0.03
0.00
0.01
-2.36
1.06
0.73
-0.44
-0.47
0.02
-0.72
-0.11
-0.18
0.38
0.03
0.09
37
38
39
40
41
42
-0.11
-0.02
0.04
-0.03
-0.08
0.01
-1.37
-0.22
0.51
-0.42
-1.06
0.08

262
056
048
247
045
056
250
218
049
115
215
040
097
198
175
047
185
182
038
210
136
100
047
276
080
041
260
073
077
071
220
046
069
257
039
043
151
181
065
152
210
032
047
244
151
040
232
125
030
049
280
083
044
282
080
031
069
230
055
063
255
042
045
152
185
049
085
161
035
205
148
042
185
107
031
246
098
046
247
053
032
294
061
074
231
040
050
242
046
103
167
039
092
181
045
200
117
034
189

300
200
z(t)
z(t)
302
107
055
237
079
035
256
100
0
Index
10
20
30
40
50
60
70
80
90
100
115
25
LBQ
15
Lag Corr
0.92 756.98
1.71 808.97
1.93 879.42
1.55 928.41
0.80 942.46
0.33
0.61
0.71
0.59
0.31
22
23
24
25
26
LBQ
-0.08 556.18
-1.19 574.07
-1.88 620.22
-2.03 678.36
-1.72 724.17
-1.05 742.21
-0.10 742.38
A
uto
co
rre
la
tio
n
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag Corr
LBQ
15 -0.03
16 -0.38
17 -0.60
18 -0.67
19 -0.59
20 -0.37
21 -0.04
-1.78 310.66
-0.18 310.88
1.54 328.22
2.82 389.72
3.10 476.22
2.42 538.84
1.20 556.10
Lag Corr
Lag Corr
LBQ
8 -0.43
9 -0.04
10 0.38
11 0.71
12 0.84
13 0.71
14 0.37
0.81 8.38 72.22

0.43 2.89 92.32
-0.03 -0.17 92.40
-0.43 -2.70 113.13
-0.69 -4.03 166.37
-0.78 -3.99 235.17
-0.68 -3.09 289.33
1
2
3
4
5
6
7

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
15
25
P
a
rtia
lA
uto
co
rre
la
tio
n
)Partial Autocorrelation Function for z(t
5
Lag
Lag PAC
PAC
0.15
0.70
-0.72
-0.23
1.01
22 0.01
23 0.07
24 -0.07
25 -0.02
26 0.10
1.93
0.84
1.10
0.39
-0.22
-0.29
0.46
15 0.19
16 0.08
17 0.11
18 0.04
19 -0.02
20 -0.03
21 0.04
Lag PAC
Lag PAC
-0.31
1.91
2.76
1.82
0.74
-0.99
-1.52
8 -0.03
9 0.19
10 0.27
11 0.18
12 0.07
13 -0.10
14 -0.15
8.38
-7.17
-2.98
-2.28
-1.61
-3.29
-1.60
0.81
-0.70
-0.29
-0.22
-0.16
-0.32
-0.16
1
2
3
4
5
6
7
:
ARIMA

.
. .
-1 : SARIMA(0,0,0)(0,1,1)12

( 1 B ) z = ( 1 B ) a
12
12
zn + l = zn + l12 + an + l an + l12
zn ( 1) = zn 11 an 11
zn ( 2 ) = zn 10 an 10
M
zn ( 12 ) = zn an
zn ( l) = zn ( l 12 ) , l 12
116

zn + l12 an + l12 ,
zn ( l) =
zn ( l 12 ) ,
l = 1, 2,...,12
l > 12
zn ( 1) = zn ( 13) = zn ( 25) = L
zn ( 2 ) = zn ( 14 ) = zn ( 26 ) = L
M
zn ( 12 ) = zn ( 24 ) = zn ( 36 ) = L
V en ( l) =
( 1 +
2
1
+ L +
2
l1
( )
1 ,
j =
0,
j = 12, 24,...
otherwise

l 1
2
V en ( l) = 2 1 +
( 1 )
12
. x x
: SARIMA(0,1,1)(0,1,1)12 -2

( 1 B ) ( 1 B12 ) zt = ( 1 B ) ( 1 B12 ) at
zn + l = zn + l1 + zn +l12 zn + l13 + an + l an +l1 an + l12 + an + l13

zn ( 1) = zn + zn 11 zn 13 an an 11 + an 12
zn ( 2 ) = zn ( 1) + zn 10 zn 11 an 10 + an 11
M
zn ( 12 ) = zn ( 11) + zn zn 1 an + an 1
zn ( 13) = zn ( 12 ) + zn ( 1) zn + an
zn ( l) = zn ( l 1) + zn ( l 12 ) zn ( l 13)
zn ( 1) = zn + zn 11 zn 13 an an 11 + an 12
zn ( 2 ) = zn ( 1) + zn 10 zn 11 an 10 + an 11
M
zn ( 12 ) = zn ( 11) + zn zn 1 an + an 1
zn ( 13) = zn ( 12 ) + zn ( 1) zn + an
zn ( l) = zn ( l 1) + zn ( l 12 ) zn ( l 13) , l > 13
117
.
:
) : (1 SARIMA :
653
736
615
688
775
845
740
800
883
947
817
566
770
653
645
696
871
783
763
773
969
858
600
705
702
667
717
805
819
800
821
902
896
582
688
756
661
698
785
859
790
805
889
937
553
618
782
697
660
690
886
834
766
778
966
577
628
722
735
687
734
824
869
810
828
903
568
598
709
798
681
711
807
913
809
802
892
599
565
622
811
722
677
707
942
855
760
782
697
583
634
736
817
701
734
860
935
784
813
640
587
658
755
767
706
750
878
894
791
834
727
617
594
635
837
740
690
756
961
837
773
656
660
611
677
784
783
723
804
900
881
812
561
742
639
602
667
858
764
711
799
957
867
797
640
716
604
635
762
826
725
751
890
924
815
843
)z(t
589
673
678
621
713
796
801
747
826
898
908
827

1000
900
)z(t
800
700
600
150
50
100
Index

) yt = ln ( zt
6.9
6.8
6.7
y
)(t
6.6
6.5
6.4
6.3
150
50
100
Index

) xt = ( 1 B ) yt = ( 1 B ) ln ( zt
118
0.15
0.10
y
(t)-y
(t-1
)
0.05
0.00
-0.05
-0.10
100
150
Index
50
.

Autocorrelation Function for y(t)-y(t
40
LBQ
30
Corr
Lag
37 0.01 0.05 713.16

38 0.20 0.90 722.06
39 -0.04 -0.19 722.45
40 -0.27 -1.21 738.86
41 -0.05 -0.24 739.51
20
Autocorrelation
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
10
LBQ
Corr
Lag
LBQ
Corr
Lag
LBQ
Corr
Lag
497.35
506.50
507.16
526.45
527.14
563.05
563.16
582.05
582.58
589.84
589.85
713.14
0.08
1.10
-0.29
-1.57
-0.29
-2.09
-0.12
-1.47
-0.24
0.89
0.02
3.63
0.02
0.21
-0.06
-0.31
-0.06
-0.42
-0.02
-0.30
-0.05
0.18
0.00
0.76
25
26
27
28
29
30
31
32
33
34
35
36
261.47
271.29
272.30
293.73
294.45
334.67
334.88
354.96
355.36
363.55
363.58
497.30
0.12
1.51
-0.47
-2.18
-0.39
-2.88
-0.20
-1.93
-0.26
1.20
0.08
4.76
0.02
0.23
-0.07
-0.34
-0.06
-0.46
-0.03
-0.32
-0.05
0.21
0.01
0.82
13
14
15
16
17
18
19
20
21
22
23
24
0.02
10.70
11.88
35.60
36.28
79.97
80.19
102.37
102.80
112.29
112.32
261.41
0.12
3.23
-1.01
-4.48
-0.68
-5.41
-0.33
-3.30
-0.43
2.01
0.11
7.74
0.01
0.25
-0.08
-0.37
-0.06
-0.50
-0.04
-0.35
-0.05
0.23
0.01
0.90
1
2
3
4
5
6
7
8
9
10
11
12
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
30
40
Lag
PAC
-0.51
0.30
0.48
0.81
-0.20
37 -0.04
38 0.02
39 0.04
40 0.06
41 -0.02
10
20
T
PAC
Lag
PAC
Lag
PAC
Lag
-1.67
-0.61
0.84
0.05
-0.94
0.34
0.93
-0.09
-1.31
0.80
0.39
0.13
-0.13
-0.05
0.06
0.00
-0.07
0.03
0.07
-0.01
-0.10
0.06
0.03
0.01
25
26
27
28
29
30
31
32
33
34
35
36
-0.48
-4.26
0.06
2.29
0.07
1.06
-1.04
0.98
0.29
-0.53
0.29
1.04
-0.04
-0.33
0.00
0.18
0.01
0.08
-0.08
0.08
0.02
-0.04
0.02
0.08
13
14
15
16
17
18
19
20
21
22
23
24
0.12
3.23
-1.20
-5.97
-0.22
-4.52
-1.82
-6.26
-5.36
-2.99
-7.55
8.08
0.01
0.25
-0.09
-0.46
-0.02
-0.35
-0.14
-0.48
-0.41
-0.23
-0.58
0.63
1
2
3
4
5
6
7
8
9
10
11
12
Partial Autocorrelation Function for y(t)-y(t
12
24 36 ) wt = ( 1 B12 ) ( 1 B ) ln ( zt

119
y(t)-y(t-1)12
0.05
0.00
-0.05
Index
50
100
150

Autocorrelation
Autocorrelation Function for y(t)-y(t

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
15
25
Corr
LBQ
Lag
Corr
LBQ
Lag
Corr
LBQ
1
2
3
4
5
6
7
8
9
10
11
12
-0.21
-0.01
0.10
-0.13
-0.10
-0.02
0.12
0.05
-0.05
0.13
-0.01
-0.44
-2.65
-0.13
1.14
-1.49
-1.11
-0.28
1.33
0.54
-0.58
1.45
-0.09
-4.90
7.15
7.17
8.63
11.20
12.67
12.77
14.95
15.31
15.75
18.49
18.50
50.88
13
14
15
16
17
18
19
20
21
22
23
24
0.18
-0.07
-0.05
0.03
0.12
-0.00
-0.11
0.03
-0.02
-0.09
0.11
-0.04
1.80
-0.70
-0.48
0.26
1.13
-0.02
-1.08
0.25
-0.20
-0.80
0.99
-0.38
56.64
57.54
57.97
58.10
60.58
60.58
62.91
63.04
63.12
64.45
66.52
66.82
25
26
27
28
29
30
31
32
33
34
35
36
0.07
-0.00
-0.06
0.03
-0.10
0.01
0.04
0.00
0.02
0.00
0.09
-0.06
0.62
-0.02
-0.57
0.30
-0.90
0.08
0.38
0.04
0.19
0.03
0.80
-0.51
67.67
67.67
68.38
68.58
70.40
70.41
70.74
70.74
70.83
70.83
72.36
72.99
Lag
35
Lag
Corr
LBQ
37 -0.06 -0.54
38 0.03 0.26
73.71
73.88
Partial Autocorrelation Function for y(t)-y(t

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
15
25
Lag
PAC
Lag
PAC
Lag
PAC
1
2
3
4
5
6
7
8
9
10
11
12
-0.21
-0.06
0.08
-0.09
-0.15
-0.10
0.11
0.11
-0.04
0.06
0.05
-0.42
-2.65
-0.73
1.06
-1.16
-1.82
-1.19
1.41
1.42
-0.51
0.80
0.59
-5.23
13
14
15
16
17
18
19
20
21
22
23
24
-0.00
-0.02
0.00
-0.11
0.04
0.01
-0.04
-0.03
-0.03
0.03
0.07
-0.27
-0.03
-0.25
0.00
-1.40
0.47
0.12
-0.52
-0.35
-0.31
0.32
0.91
-3.41
25
26
27
28
29
30
31
32
33
34
35
36
0.12
-0.01
-0.12
-0.07
0.04
-0.04
-0.05
0.02
-0.01
-0.02
0.18
-0.21
1.51
-0.15
-1.44
-0.91
0.44
-0.53
-0.60
0.19
-0.12
-0.19
2.21
-2.57
35
Lag
PAC
37 0.10
38 -0.06
1.23
-0.79
wt = ( 1 B12 ) ( 1 B ) ln ( zt )
0 0 q p
( 1 B ) ( 1 B ) ln ( z ) = ( 1 B ) a
12
12
MINITAB SARIMA(0,1,0)(0,1,1)12
120
ARIMA 0 1 0 0 1 1 12 'y(t)' ;
NoConstant.
zt = e y yt = ln ( zt )
:
t
MTB > Name c14 = 'RESI3' c15 = 'FITS3'

MTB > ARIMA 0 1 0 0 1 1 12 'y(t)' 'RESI3' 'FITS3';
SUBC>
NoConstant;
SUBC>
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
SUBC> GNormalplot.
ARIMA Model
ARIMA model for y(t)
Iteration
SSE
Parameters
0
0.0228597
0.100
1
0.0204943
0.250
2
0.0187066
0.400
3
0.0174234
0.550
4
0.0169841
0.684
5
0.0169841
0.683
6
0.0169841
0.683
Type
Coef
StDev
SMA 12
0.6831
0.0610
0.0010
T
11.20
Differencing: 1 regular, 1 seasonal of order 12

Residuals:
MS = 0.0001077 DF = 154
Lag
12
24
36
Chi-Square
9.0(DF=11)
29.9(DF=23)
44.5(DF=35)
48
59.4(DF=47)

Period
169
170
171
172
173
174
175
176
177
178
179
180
181
182
Forecast
6.76750
6.70901
6.83815
6.85381
6.92288
6.89349
6.84654
6.80008
6.74395
6.75028
6.70664
6.75999
6.79052
6.73203
95 Percent Limits
Lower
Upper
6.74716
6.78784
6.68024
6.73778
6.80292
6.87338
6.81313
6.89450
6.87739
6.96836
6.84366
6.94331
6.79272
6.90035
6.74255
6.85761
6.68293
6.80497
6.68596
6.81461
6.63918
6.77410
6.68952
6.83045
6.71514
6.86590
6.65203
6.81203
121
Actual
183
184
185
186
187
188
189
190
191
192
6.86117
6.87684
6.94590
6.91651
6.86956
6.82310
6.76697
6.77330
6.72966
6.78301
6.77680
6.78832
6.85342
6.82023
6.76962
6.71963
6.66009
6.66312
6.61627
6.66649
( 1 B ) ( 1 B ) ln ( z ) = ( 1 0.683B ) a ,
12
12
6.94554
6.96535
7.03838
7.01279
6.96950
6.92657
6.87385
6.88349
6.84305
6.89952
at : N ( 0,0.0001077 )
= 0.683, s.e. ( ) = 0.061, t = 11.2
.
:

MTB > ZTest 0.0 0.0103778 'RESI3';
SUBC>
Alternative 0.
Z-Test
Variable
RESI3
N
Mean
StDev
155 -0.000111 0.010375
SE Mean
0.000834
Z
-0.13
P
0.89
0.05 P-value=0.89

Runs Test
RESI3
K =
0.0000

83 below
0.1893
:

122
ACF of Residuals for y(t)

1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
3
12
15
18
21
24
27
30
33
36
39
33
36
39
Lag
PACF of Residuals for y(t)

1.0
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
3
12
15
18
21
24
27
30
Lag
.
:
(response is y(t))
30
Frequency
0.8
20
10
0
-0.03
-0.02
-0.01
0.00
0.01
0.02
Residual
123
0.03
0.04
Normal Probability Plot of the Residuals

))(response is y(t
0.04
0.03
0.02
0.00
Residual
0.01
-0.01
-0.02
-0.03
3
-1
-3
-2
Normal Score
K-S test for Residuals
.999
.99
.95
.50
.20
Probability
.80
.05
.01
.001
0.03
0.02
0.01
-0.01
0.00
-0.02
-0.03
RESI3
Average: -0.0001115
StDev: 0.0103754
N: 155

D+: 0.074 D-: 0.045 D : 0.074
P-value K-S 0.041 0.05

= 0.05 .
:
24 95%
:
1150
1050
850
750
25
20
10
15
Time
124
Forecast
950
1150
1050
850
750
Forecast
950
650
550
200
100
Time
: 1
SARIMA :
57.4
60.0
69.8
70.1
66.1
58.7
55.8
51.6
53.5
53.0
52.1
53.6
55.6
62.6
75.5
76.4
61.5
71.0
71.0
67.5
56.1
55.4
55.3
52.4
53.4
53.5
53.4
55.3
60.1
70.3
73.3
70.8
72.7
70.6
67.4
58.0
53.6
52.9
53.2
52.1
53.6
54.3
55.3
58.5
68.9
69.4
68.1
60.6
72.2
68.2
58.2
54.2
53.0
53.0
52.8
52.6
55.1
56.3
58.2
69.3
68.8
64.7
58.1
56.4
71.5
57.7
54.3
53.2
52.8
52.7
52.8
53.9
57.3
59.4
66.9
69.6
63.6
57.1
54.8
55.6
59.1
54.6
53.4
53.0
52.6
53.4
54.5
56.4
61.9
70.4
70.7
64.2
57.1
53.9
54.3
55.0
57.2
54.9
53.0
53.0
52.9
54.9
56.4
61.7
69.9
69.9
65.3
55.5
52.2
53.5
54.6
54.7
56.3
54.9
52.8
53.4
54.0
57.5
59.3
68.3
71.5
61.0
56.5
53.3
51.5
53.5
54.2
55.8
55.8
54.9
53.3
54.6
54.9
61.2
68.7
67.9
65.1
52.7
53.4
52.4
51.7
53.1
54.8
57.7
55.7
55.3
54.4
55.6
56.6
69.3
70.0
65.3
57.3
53.9
52.5
51.5
51.5
53.3
55.8
60.5
62.2
70
z
)(t
60
50
150
50
100
125
Index
)z(t
56.3
56.9
56.0
59.4
68.2
69.8
67.9
58.2
52.7
53.3
53.2
51.5
52.3
53.9
57.9
66.4
74.8
wt = ( 1 B ) zt
y
(t)
10
-10
Index
50
100
150

Autocorrelation
Autocorrelation Function for y(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag
Corr
LBQ
Lag
Corr
LBQ
Lag
Corr
LBQ
Lag
Corr
LBQ
1
2
3
4
5
6
7
8
9
10
11
12
0.47
-0.02
-0.37
-0.28
-0.21
-0.19
-0.21
-0.27
-0.32
-0.00
0.46
0.86
6.32
-0.22
-4.10
-2.82
-2.00
-1.77
-1.92
-2.46
-2.85
-0.02
3.91
6.71
40.60
40.67
65.83
80.10
87.92
94.41
102.29
115.89
135.47
135.48
176.37
318.37
13
14
15
16
17
18
19
20
21
22
23
24
0.43
-0.01
-0.33
-0.25
-0.19
-0.17
-0.19
-0.25
-0.31
-0.01
0.42
0.79
2.76
-0.03
-2.00
-1.49
-1.12
-0.97
-1.09
-1.42
-1.76
-0.06
2.35
4.28
354.69
354.69
375.77
388.04
395.15
400.68
407.71
419.89
439.21
439.24
475.10
602.75
25
26
27
28
29
30
31
32
33
34
35
36
0.41
-0.00
-0.29
-0.23
-0.17
-0.16
-0.17
-0.23
-0.28
-0.01
0.37
0.72
2.06
-0.01
-1.43
-1.08
-0.82
-0.78
-0.83
-1.06
-1.31
-0.06
1.70
3.27
638.60
638.60
656.89
667.77
674.14
679.93
686.56
697.63
714.89
714.93
745.11
860.42
37
38
39
40
41
42
43
44
0.40
0.02
-0.27
-0.21
-0.17
-0.15
-0.16
-0.20
1.72
0.07
-1.14
-0.89
-0.71
-0.61
-0.65
-0.83
896.48
896.54
913.29
923.75
930.47
935.55
941.37
950.82
12
22
32
42
Partial Autocorrelation Function for y(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
Lag PAC
1
2
3
4
5
6
7
8
9
10
11
12
0.47
-0.32
-0.30
0.09
-0.23
-0.25
-0.16
-0.43
-0.64
-0.31
-0.16
0.41
12
T
6.32
-4.21
-4.00
1.22
-3.05
-3.33
-2.09
-5.67
-8.56
-4.11
-2.18
5.42
Lag PAC
13
14
15
16
17
18
19
20
21
22
23
24
-0.27
-0.01
0.01
-0.07
-0.01
0.02
0.03
0.07
-0.12
-0.02
-0.06
0.13
22
T
-3.62
-0.17
0.13
-0.95
-0.16
0.24
0.44
0.89
-1.62
-0.26
-0.86
1.74
32
Lag PAC
25
26
27
28
29
30
31
32
33
34
35
36
-0.02
-0.06
0.07
-0.00
-0.02
-0.04
0.01
0.03
-0.02
0.00
-0.09
0.05
T
-0.33
-0.74
0.99
-0.05
-0.26
-0.60
0.20
0.37
-0.25
0.06
-1.17
0.61
42
Lag PAC
37
38
39
40
41
42
43
44
0.03
-0.03
-0.01
0.02
-0.04
0.03
-0.00
0.03
T
0.38
-0.35
-0.17
0.23
-0.56
0.35
-0.00
0.43
12
wt = ( 1 B ) ( 1 B ) zt 12
126
w(t)
-5
Index
50
100
150

Autocorrelation

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
10
20
30
Corr
LBQ
Lag
Corr
LBQ
Lag
Corr
LBQ
1
2
3
4
5
6
7
8
9
10
11
12
-0.01
-0.19
-0.24
-0.00
0.00
-0.04
-0.04
-0.03
0.22
0.10
0.08
-0.42
-0.19
-2.48
-2.97
-0.03
0.03
-0.50
-0.49
-0.31
2.56
1.16
0.93
-4.63
0.04
6.31
16.10
16.10
16.10
16.41
16.71
16.83
25.26
27.13
28.37
59.40
13
14
15
16
17
18
19
20
21
22
23
24
-0.08
0.10
0.12
-0.02
-0.03
0.10
0.07
-0.01
-0.20
-0.02
0.10
0.14
-0.84
0.98
1.15
-0.17
-0.28
1.01
0.70
-0.05
-1.94
-0.23
0.95
1.30
60.69
62.50
65.01
65.07
65.22
67.24
68.23
68.24
75.94
76.05
78.02
81.78
25
26
27
28
29
30
31
32
33
34
35
36
-0.08
-0.13
-0.01
0.06
0.12
-0.15
-0.08
-0.03
0.19
0.17
-0.10
-0.22
-0.76
-1.18
-0.13
0.56
1.09
-1.36
-0.72
-0.28
1.66
1.49
-0.86
-1.92
83.09
86.28
86.33
87.07
89.94
94.47
95.80
96.01
103.20
109.19
111.24
121.77
Lag
40
Lag
LBQ
37 0.05 0.46
38 0.10 0.84
39 0.02 0.15
40 -0.02 -0.16
41 -0.12 -1.01
Corr
122.40
124.52
124.59
124.67
127.84

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
10
20
30
Lag
PAC
Lag
PAC
Lag
PAC
1
2
3
4
5
6
7
8
9
10
11
12
-0.01
-0.19
-0.26
-0.07
-0.11
-0.14
-0.11
-0.13
0.14
0.07
0.17
-0.32
-0.19
-2.48
-3.28
-0.85
-1.43
-1.84
-1.46
-1.66
1.76
0.95
2.19
-4.08
13
14
15
16
17
18
19
20
21
22
23
24
-0.05
0.01
-0.04
-0.03
-0.02
0.06
0.05
-0.03
-0.05
0.05
0.19
0.01
-0.69
0.15
-0.50
-0.33
-0.23
0.78
0.62
-0.42
-0.68
0.63
2.50
0.09
25
26
27
28
29
30
31
32
33
34
35
36
-0.12
-0.08
-0.03
-0.03
0.07
-0.11
-0.02
-0.10
-0.01
0.18
0.07
-0.09
-1.49
-0.97
-0.32
-0.36
0.87
-1.39
-0.29
-1.29
-0.09
2.31
0.95
-1.18
Lag
40
PAC
37 0.02
38 -0.08
39 0.03
40 0.05
41 -0.03
0.24
-1.06
0.44
0.63
-0.36

SARIMA(1,1,1)(0,1,1)12
127
( 1 B ) ( 1 B12 ) ( 1 B ) zt = ( 1 B ) ( 1 B12 ) at
:
MTB > ARIMA 1 1 1 0 1 1 12 'z(t)' 'RESI2';
SUBC>
NoConstant;
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
SUBC> GNormalplot.
ARIMA Model
Iteration
SSE
Parameters
0
307.653
0.100
0.100
1
281.217
0.100
0.100
2
262.275
0.226
0.231
3
262.027
0.376
0.381
4
261.770
0.526
0.531
5
261.426
0.675
0.681
6
260.905
0.824
0.831
7
260.036
0.970
0.981
8
227.926
0.835
0.980
9
221.838
0.748
0.980
10
221.665
0.738
0.980
11
221.637
0.738
0.980
12
221.610
0.737
0.980
13
221.585
0.737
0.980
Type
Coef
StDev
AR
1
0.7374
0.0620
MA
1
0.9796
0.0017
SMA 12
0.5898
0.0736
0.100
0.250
0.400
0.401
0.401
0.402
0.403
0.405
0.536
0.576
0.586
0.589
0.589
0.590
0.0010
T
11.89
582.86
8.01
Differencing: 1 regular, 1 seasonal of order 12

Residuals:
MS =
1.323 DF = 162
Lag
12
24
36
Chi-Square
15.7(DF= 9)
30.9(DF=21)
61.6(DF=33)
48
67.1(DF=45)
( 1 0.74 B ) ( 1 B ) ( 1 B ) zt = ( 1 0.98B ) ( 1 0.59 B12 ) at ,

12
at : N ( 0,1.323)
= 0.74, s.e. ( ) = 0.062, t = 11.89
= 0.96, s.e. ( ) = 0.0017, t = 582.86

= 0.59, s.e. ( ) = 0.074, t = 8.01
.
:
128

SUBC>
Alternative 0.
Z-Test
Variable
RESI1
N
165
Mean
-0.0144
StDev
1.1433
SE Mean
0.0895
Z
-0.16
P
0.87
0.05 P-value=0.87

Runs Test
RESI1
K =
0.0000

92 below
0.05
: Sign Test
MTB > STest 0.0 'RESI1';
SUBC>
Alternative 0.
Sign Test for Median

Sign test of median = 0.00000 versus
RESI1
N
165
N*
13
Below
92
Equal
0
not =
Above
73
0.00000
P
0.1611
Median
-0.08139
0.1611
MTB > WTest 0.0 'RESI1';

SUBC>
Alternative 0.
Wilcoxon Signed Rank Test

Test of median = 0.000000 versus median not = 0.000000
RESI1
N
165
Number
Missing
13
N for
Test
165
Wilcoxon
Statistic
6321.0
P
0.392
Estimated
Median
-0.05940
0.392
:
129

1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
3
12
15
18
21
24
27
30
33
36
39
36
39
Lag

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
3
12
15
18
21
24
27
30
33
Lag
.
:
(response is z(t))
50
Frequency
40
30
20
10
0
-6
-5
-4
-3
-2
-1
Residual
130

(response is z(t))
4
3
2
Residual
1
0
-1
-2
-3
-4
-5
-3
-2
-1
Normal Score
K-S Test for Residuals

.999
Probability
.99
.95
.80
.50
.20
.05
.01
.001
-5
-4
-3
-2
-1
RESI1
Average: -0.0144171
StDev: 1.14327
N: 165

D+: 0.117 D-: 0.140 D : 0.140
Approximate P-Value < 0.01
= 0.05 0.01 K-S P-value

.
:
95% 36
Period
179
180
181
182
183
184
185
186
187
188
189
190
191
Forecast
57.7885
55.8516
54.7429
54.1820
54.6298
54.9152
55.6640
59.3778
68.6924
74.0698
73.9650
63.5866
58.9095
95 Percent Limits
Lower
Upper
55.5332
60.0437
53.0220
58.6812
51.6264
57.8594
50.9063
57.4578
51.2601
57.9994
51.4875
58.3430
52.1986
59.1294
55.8869
62.8688
65.1833
72.2015
70.5472
77.5925
70.4317
77.4983
60.0447
67.1286
55.1843
62.6347
131
Actual
52.9407
51.6169
50.9022
51.2380
51.4409
52.1278
55.7947
65.0729
70.4217
70.2940
59.8968
55.0418
52.7962
51.4676
50.7472
51.0774
51.2749
51.9568
55.6189
64.8927
70.2374
70.1057
59.7046
60.6128
59.4305
58.8105
59.2131
59.4653
60.1905
63.8885
73.1929
78.5647
78.4575
68.0793
63.5664
61.5364
60.3514
59.7315
60.1357
60.3903
61.1184
64.8194
74.1270
79.5021
79.3983
69.0235
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
56.7768
55.5237
54.8564
55.2256
55.4531
56.1592
59.8416
69.1329
74.4932
74.3758
63.9881
59.3041
57.1663
55.9095
55.2394
55.6066
55.8326
56.5376
60.2191
69.5099
74.8697
74.7520
64.3640
:
80
60
F
o
re
c
a
s
t
70
50
30
40
20
10
Time
80
60
50
200
100
Time
132
Forecast
70
133
Forecasting By ARMA Models

28.0
28.0
29.5
23.5
30.5
32.3
21.3
29.6
30.6
24.5
36.5
26.5
23.5
36.5
20.7
16.4
15.3
25.0
16.0
23.4
15.5
25.3
19.0
26.4
21.0
17.3
19.7
32.2
12.0
24.0
26.0
28.3
20.5
15.5
21.5
31.3
:
;MTB > TSPlot C1
>SUBC
;Index
>SUBC
;TDisplay 11
>SUBC
;Symbol
>SUBC
Connect.
30
C
1
20
10
30
20
134
10
Index
MTB > %ACF C1.
Autocorrelation
Autocorrelation Function for C1

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag
Corr
LBQ
1
2
3
4
5
6
7
0.63
0.30
0.14
-0.05
-0.24
-0.30
-0.24
3.79
1.35
0.59
-0.20
-1.01
-1.22
-0.94
15.62
19.27
20.07
20.17
22.71
26.71
29.40
Lag
Corr
LBQ
8 -0.20 -0.78
9 -0.12 -0.44
31.42
32.10
MTB > %PACF C1.

Executing from file: E:\MTBWIN\MACROS\PACF.MAC
Partial Autocorrelation Function for C1

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag PAC
1
2
3
4
5
6
7
0.63
-0.16
0.04
-0.20
-0.18
-0.04
0.02
Lag PAC
3.79
-0.98
0.22
-1.20
-1.09
-0.23
0.12
8 -0.07
9 0.05
-0.39
0.29
ARMA ( 1,1)

135
MTB > ARIMA 1 0 1 C1 'RESI1';

SUBC>
Constant;
SUBC>
SUBC> GACF;
SUBC> GPACF;
SUBC> GNormalplot.
ARIMA Model
ARIMA model for C1
Iteration
SSE
Parameters
0
1337.71
0.100
0.100
21.918
1
936.95
0.250
-0.049
18.193
2
849.78
0.211
-0.199
19.106
3
751.53
0.215
-0.349
18.941
4
658.66
0.266
-0.499
17.594
5
592.30
0.372
-0.649
14.890
6
580.80
0.433
-0.699
13.314
7
579.30
0.455
-0.714
12.698
8
579.11
0.464
-0.719
12.470
9
579.08
0.467
-0.721
12.386
10
579.08
0.468
-0.722
12.356
11
579.08
0.468
-0.722
12.345
Relative change in each estimate less than 0.0010
Type
Coef
StDev
AR
1
0.4684
0.1755
MA
1
-0.7221
0.1380
Constant
12.345
1.154
Mean
23.221
2.170
T
2.67
-5.23
10.70
Residuals:
MS = 15.860 DF = 33
Lag
12
24
36
48
Chi-Square 7.2(DF=10)
15.9(DF=22) * (DF= *)
* (DF= *)
Period
37
38
39
40
41
Forecast
14.7649
19.2606
21.3663
22.3524
22.8143
95 Percent Limits
Lower
Upper
6.9578
22.5720
7.1228
31.3985
8.4715
34.2610
9.2975
35.4074
9.7245
35.9041
Actual
:
zt = 12.345 + 0.4684 zt 1 + at 0.7221at 1 , at : WN ( 0,15.86 ) t
136

1 = 0.4684 se 1 = 0.1755 t = 2.67
) (
= 0.7221 se ( ) = 0.1380 t = 5.23
= 12.345 se ( ) = 1.154 t = 10.70
1
= 0.05
2 = 15.86 df = 33
H 0 : 1 = 0
H1 : 1 0
0.4684
= 2.6689
0.1755
1 = 0
) (
se 1
=t
= 0.05
H 0 : a = 0, H 1 : a 0
;'MTB > TTest 0.0 'RESI1

>SUBC
Alternative 0.
T-Test of the Mean

P
0.60
T
0.54
SE Mean
0.642
StDev
3.851
Mean
0.344
N
36
Variable
RESI1
0.6 = 0.05
P-Value
t = 0.54

:
Runs Test
MTB > Runs 'RESI1'.
Runs Test
RESI1
0.3443
= K

18 below
137
= 0.05
:

ACF of Residuals for C1
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
1
Lag
PACF of Residuals for C1

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
1
Lag

138

(response is C1)
Residual
10
-10
-2
-1
Normal Score
( )

95%
MTB > TSPlot C14 C15 C16;

SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect;
SUBC> overlay.
Time Series Plot for C1

(with forecasts and their 95% confidence limits)
36
C1
26
16
6
5
10
15
20
25
30
35
Time
:
10.38
11.68
9.10
10.01
9.51
9.48
11.86
11.17
9.09
9.37
9.24
7.38
10.97
10.53
9.35
8.69
8.66
6.90
10.80
9.79 10.39 10.42 10.82 11.40
10.01
9.91
9.14
9.16
9.55
9.67
8.82
9.32
9.01
9.00
9.80
9.83
8.19
8.67
9.55
8.92
8.09
9.37
8.86
8.05
7.79
6.75
6.75
7.82
6.94
6.24
6.84
6.85
6.90
7.79
139
11.32 11.44
8.44
8.24
9.72
9.89
10.13 10.14
8.64 10.58
8.18
7.51
7.23
10.85
6.89
8.42
10.41
5.96
9.61
9.96
6.80
9.05
9.61
7.68
9.26
8.76
8.38
9.22
8.18
8.52
9.38
7.21
9.74
9.10
7.13
9.31
7.95
9.10
9.89
8.12
8.25
9.96
9.75
7.91
MTB > TSPlot C10;

SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect.
12
11
C
1
0
10
9
8
7
6
Index
10
20
30
40
50
60
70
80
90
Autocorrelation
Autocorrelation Function for C10

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
Lag Corr
1
2
3
4
5
6
7
0.83
0.61
0.46
0.37
0.33
0.28
0.26
12
LBQ
8.24 69.92
3.91 107.90
2.56 129.56
1.95 143.87
1.65 155.04
1.40 163.68
1.28 171.23
Lag Corr
8
9
10
11
12
13
14
0.26
0.26
0.18
0.09
0.04
0.03
0.04
LBQ
1.26 178.83
1.21 186.14
0.84 189.86
0.43 190.87
0.20 191.09
0.13 191.19
0.19 191.39
22
Lag Corr
LBQ
15 0.05 0.21 191.63

16 0.04 0.16 191.78
17 0.00 0.02 191.78
18 -0.03 -0.15 191.91
19 -0.05 -0.24 192.26
20 -0.05 -0.24 192.60
21 0.01 0.07 192.63
Lag Corr
Partial Autocorrelation Function for C10

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
12
Lag PAC
Lag PAC
1 0.83
2 -0.27
3 0.13
4 0.03
5 0.06
6 -0.02
7 0.09
8.24
-2.64
1.29
0.34
0.61
-0.21
0.91
8 0.05
9 0.00
10 -0.20
11 0.02
12 0.01
13 0.01
14 0.03
0.45
0.03
-1.98
0.19
0.09
0.12
0.34
22
Lag PAC
140
15
16
17
18
19
20
21
-0.01
-0.03
-0.07
-0.03
0.06
0.02
0.21
LBQ
22 0.10 0.47 194.02

23 0.18 0.81 198.12
24 0.20 0.89 203.24
Lag PAC
-0.15
-0.25
-0.73
-0.26
0.60
0.20
2.03
22 0.05
23 0.06
24 -0.07
0.51
0.59
-0.65
AR ( 2 )

MTB > ARIMA 2 0 0 C10 'RESI1';
SUBC>
Constant;
SUBC>
SUBC> GSeries;
SUBC> GACF;
SUBC> GPACF;
SUBC> GNormalplot.
ARIMA Model
ARIMA model for C10
Iteration
SSE
Parameters
0
126.398
0.100
0.100
1
103.515
0.250
0.043
2
84.535
0.400
-0.014
3
69.407
0.550
-0.071
4
58.132
0.700
-0.128
5
50.724
0.850
-0.184
6
47.212
1.000
-0.239
7
46.918
1.053
-0.256
8
46.916
1.054
-0.255
9
46.916
1.054
-0.255
10
46.916
1.054
-0.255
Type
Coef
StDev
AR
1
1.0542
0.0992
AR
2
-0.2547
0.0993
Constant
1.81360
0.07092
Mean
9.0480
0.3538
7.283
6.434
5.586
4.738
3.887
3.030
2.163
1.838
1.816
1.814
1.814
0.0010
T
10.63
-2.56
25.57
Residuals:
MS = 0.4921 DF = 95
Lag
12
24
36
48
Chi-Square 7.2(DF=10) 13.7(DF=22) 21.3(DF=34) 28.8(DF=46)
141
Period
99
100
101
102
103
Forecast
9.7950
9.6033
9.4432
9.3232
9.2375
95 Percent Limits
Lower
Upper
8.4198
11.1703
7.6050
11.6016
7.1234
11.7629
6.8446
11.8018
6.6825
11.7925
Actual

zt = 1.8136 + 1.0542 zt 1 0.2547 zt 2 + at , at : WN ( 0,0.4921) t
t
1 = 1.0542 se 1 = 0.0992 t = 10.63
( )
= 0.2547 se ( ) = 0.0993 t = 2.56
= 1.8136 se ( ) = 0.07092 t = 25.57
2
2 = 0.4921 df = 95
= 0.05

MTB > TTest 0.0 'RESI1';

SUBC>
Alternative 0.
T-Test of the Mean

Variable
RESI1
N
98
Mean
StDev
-0.0082 0.6942
SE Mean
0.0701
T
-0.12
P
0.91
P-Value )
(

MTB > Runs 'RESI1'.
Runs Test
RESI1
K =
-0.0082

51 below

142
ACF of Residuals for C10

1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
10
12
14
16
18
20
22
24
22
24
Lag
PACF of Residuals for C10

1.0
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
2
10
12
14
16
18
20
Lag
(response is C10)
2
Residual
0.8
-1
-2
-3
-2
-1
143 0
Normal Score
) )( at : IIDN ( 0,0.4921
95%
;MTB > TSPlot C14 C15 C16
>SUBC
;Index
>SUBC
;TDisplay 11
>SUBC
;Symbol
>SUBC
;Connect
SUBC> overlay.
Time Series Plot for C10

)(with forecasts and their 95% confidence limits
12
11
10
8
7
6
90
80
70
60
50
40
30
20
10
Time
: ) AR ( 1
144
C10
145

Example on Residual Analysis and Model Selection
: Criteria

z1 , z2 ,..., zn z1 , z2 ,..., zn :
ei = zi zi , i = 1,2,..., n
ai = ei , i = 1,2,..., n
:
-1
-2 )
2
2 ) ( at : IIDN ( 0,

:
H 0 : E ( at ) = 0
H 1 : E ( at ) 0
= u

) se ( e
= 0.05 E ( at ) = 0 ) u < 1.96
30 (
:
Metals
MTB > RETR 'E:\Mtbwin\DATA\EMPLOY.MTW'.

Retrieving worksheet from file: E:\Mtbwin\DATA\EMPLOY.MTW
Worksheet was saved on 6/ 5/1996
;'MTB > TSPlot 'Metals
>SUBC
;Index
>SUBC
;TDisplay 11
>SUBC
;Symbol
>SUBC
Connect.
50
M
e
ta
ls
45
40
60
50
40
30
20
10
Index
'MTB > Name c4 = 'AVER1' c5 = 'FITS1' c6 = 'RESI1
146
MTB > %MA 'Metals' 3;

SUBC>
Averages 'AVER1';
SUBC>
Fits 'FITS1';
SUBC>
Residuals 'RESI1'.
Executing from file: E:\MTBWIN\MACROS\MA.MAC
Moving average
Data
Length
NMissing
Metals
60.0000
0
Moving Average
Length: 3
Accuracy Measures
MAPE: 1.55036
MAD: 0.70292
MSD: 0.76433
Moving Average
Actual
Predicted
M
etals
50
Actual
Predicted
45
Moving Average
Length:
40
0
10
20
30
40
50
MAPE:
1.55036
MAD:
0.70292
MSD:
0.76433
60
Time

MTB > print c3 c6 c5
Data Display
Row
Metals
RESI1
FITS1
1
2
3
4
5
6
7
8
9
44.2
44.3
44.4
43.4
42.8
44.3
44.4
44.8
44.4
*
*
*
-0.90000
-1.23333
0.76667
0.90000
0.96667
-0.10000
*
*
*
44.3000
44.0333
43.5333
43.5000
43.8333
44.5000
147
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
43.1
42.6
42.4
42.2
41.8
40.1
42.0
42.4
43.1
42.4
43.1
43.2
42.8
43.0
42.8
42.5
42.6
42.3
42.9
43.6
44.7
44.5
45.0
44.8
44.9
45.2
45.2
45.0
45.5
46.2
46.8
47.5
48.3
48.3
49.1
48.9
49.4
50.0
50.0
49.6
49.9
49.6
50.7
50.7
50.9
50.5
51.2
50.7
50.3
49.2
48.1
-1.43333
-1.50000
-0.96667
-0.50000
-0.60000
-2.03333
0.63333
1.10000
1.60000
-0.10000
0.46667
0.33333
-0.10000
-0.03333
-0.20000
-0.36667
-0.16667
-0.33333
0.43333
1.00000
1.76667
0.76667
0.73333
0.06667
0.13333
0.30000
0.23333
-0.10000
0.36667
0.96667
1.23333
1.33333
1.46667
0.76667
1.06667
0.33333
0.63333
0.86667
0.56667
-0.20000
0.03333
-0.23333
1.00000
0.63333
0.56667
-0.26667
0.50000
-0.16667
-0.50000
-1.53333
-1.96667
44.5333
44.1000
43.3667
42.7000
42.4000
42.1333
41.3667
41.3000
41.5000
42.5000
42.6333
42.8667
42.9000
43.0333
43.0000
42.8667
42.7667
42.6333
42.4667
42.6000
42.9333
43.7333
44.2667
44.7333
44.7667
44.9000
44.9667
45.1000
45.1333
45.2333
45.5667
46.1667
46.8333
47.5333
48.0333
48.5667
48.7667
49.1333
49.4333
49.8000
49.8667
49.8333
49.7000
50.0667
50.3333
50.7667
50.7000
50.8667
50.8000
50.7333
50.0667

MTB > TTest 0.0 'RESI1';
SUBC>
Alternative 0.
T-Test of the Mean
148

Variable
RESI1
N
57
Mean
0.158
StDev
0.868
SE Mean
0.115
T
1.37
P
0.17
) ( Minitab :
1.96 T=1.37 . Ttest

Runs test :
:
MTB > Runs 'RESI1'.
Runs Test
RESI1
K =
0.1579

27 below
Runs Test
RESI1
K =
0.0000

24 below

Autocorrelation test :
:
MTB > %ACF 'RESI1'.
Executing from file: E:\MTBWIN\MACROS\ACF.MAC
149
Autocorrelation
Autocorrelation Function for RESI1

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
14
Lag
Corr
LBQ
Lag
Corr
LBQ
1
2
3
4
5
6
7
0.56
0.24
-0.01
0.04
0.02
-0.07
-0.14
4.24
1.39
-0.06
0.21
0.11
-0.38
-0.81
18.93
22.31
22.32
22.40
22.42
22.72
24.07
8
9
10
11
12
13
14
-0.04
0.12
0.29
0.24
0.20
0.05
0.03
-0.22
0.68
1.64
1.30
1.05
0.25
0.17
24.17
25.21
31.36
35.67
38.71
38.90
38.99
1 = 0 4.24 T

H 0 : 1 = 0
H1 : 1 0
r1
= 4.24
se ( r1 )
:
:

SUBC>
Kstest.
Executing from file: E:\MTBWIN\MACROS\NormPlot.MAC
.999
P
ro
b
a
b
ility
.99
.95
.80
.50
.20
.05
.01
.001
-2
-1
RESI1
Average: 0.157895
StDev: 0.867525
N: 57

D+: 0.054 D-: 0.084 D : 0.084
Approximate P-Value > 0.15
0.05 0.15 P-Value

Q-Q Plot = 0.05

MTB > %Qqplot 'RESI1';
SUBC>
Table;
SUBC>
Conf 95;
SUBC>
Ci.
Executing from file: E:\MTBWIN\MACROS\Qqplot.MAC
150
Distribution Function Analysis

Normal Dist. Parameter Estimates
: RESI1
0.157895
0.867525
Data
Mean:
StDev:
0.157895
Mean:
0.867525
StDev:
99
95
90
80
60
50
40
Percent
70
30
20
10
5
1
2
-1
-2
Data

:

.
151

: Decomposition Method

. z1 , z2 ,..., zn

zt = Tt + St + Ct + Et , t = 1, 2,..., n
Zt Tt
St ) (
Ct
) ( Et

. Additive
Model .
zt = Tt St + Ct + Et , t = 1,2,..., n
zt = Tt St Ct + Et , t = 1, 2,..., n
. Multiplicative Models
Ct

.

zt = Tt + St + Et , t = 1, 2,..., n
zt = Tt St + Et , t = 1,2,..., n
FORECASTING: METHODS AND APPLICATIONS
MAKRIDAKIS/ WHEELWRIGHT/ McGEE 141-131

zt = Tt + St + Et , t = 1, 2,..., n
zt = Tt St + Et , t = 1,2,..., n
1960
1975
GasDemand
MONTHLY GASOLINE DEMAND ONTARIO GALLON MILLIONS 1960-1975
87695 86890 96442 98133 113615 123924 128924 134775 117357 114626
107677 108087 92188 88591 98683 99207 125485 124677 132543 140735
124008 121194 111634 111565 101007 94228 104255 106922 130621 125251
140318 146174 122318 128770 117518 115492 108497 100482 106140 118581
132371 132042 151938 150997 130931 137018 121271 123548 109894 106061
112539 125745 136251 140892 158390 148314 144148 140138 124075 136485
109895 109044 122499 124264 142296 150693 163331 165837 151731 142491
140229 140463 116963 118049 137869 127392 154166 160227 165869 173522
155828 153771 143963 143898 124046 121260 138870 129782 162312 167211
172897 189689 166496 160754 155582 145936 139625 137361 138963 155301
152
172026
148932
154277
174176
198688
218099
227443
193522
199024
217775
165004
156751
144998
184416
190474
229001
233038
212870
191813
227621
185861
177998
159644
158167
194502
203200
234119
248565
195997
190270
174559
168646
156261
190755
212557
255133
221532
208684
163903
198079
166273
176353
166286
197095
216478
252642
244113
174270
189073
190176
175720
170699
193693
232868
255007
243108
160272
175702
205541
193939
181468
188992
221616
206826
255918
165614
180097
193657
201269
174241
175347
209893
233231
244642
146182
155202
182617
218960
210802
196265
194784
212678
237579
137728
174508
189614
209861
212262
203526
189756
217173
237579
MTB > TSPlot 'GasDemand';

SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect.
G
asD
em
and
250
200
150
100
Index
50
100
150

zt = Tt + St + Et , t = 1, 2,..., n : :
SUBC>
SUBC>
SUBC>
SUBC>
;MTB > %Decomp 'GasDemand' 12

Additive ;
Forecasts 24;
Title "Forecast of Gasoline Demand";
Start 1.
Time Series Decomposition

Data
Length
NMissing
GasDeman
192.000
0
Trend Line Equation

Yt = 96.4074 + 0.680579*t
Seasonal Indices
Period
1
2
3
Index
-20.5625
-26.8125
-14.8958
153
4
5
6
7
8
9
10
11
12
-11.0625
9.89583
11.8958
22.7708
25.1875
5.64583
7.27083
-4.81250
-4.52083
Accuracy of Model
MAPE:
MAD:
MSD:
3.6952
5.6622
52.7851
Forecasts
Row
Period
Forecast
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
207.197
201.627
214.225
218.738
240.377
243.058
254.614
257.711
238.850
241.155
229.753
230.725
215.364
209.794
222.391
226.905
248.544
251.225
262.780
265.878
247.017
249.322
237.919
238.892
Forecast of Gasoline Demand
30
Seasonal Indices
Original Data, by Seasonal Period

250
20
10
200
0
150
-10
-20
-30
1 2 3 4 5 6 7 8 9 10 11 12
Percent Variation, by Seasonal Period
1 2 3 4 5 6 7 8 9 10 11 12
30
Residuals, by Seasonal Period
20
10
(1)
100
10
154 -100
-20
-30
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 7 8 9 10 11 12
(2)
Original Data
200
150
100
0
100
200
Seasonally Adjusted Data
250
Detrended Data
50
40
30
20
10
0
-10
-20
-30
-40
250
30
100
200
Seasonally Adj. and Detrended Data
20
200
10
150
-10
0
-20
100
-30
0
100
200
100
200
(3)
Actual
Predicted
250
GasDeman
Forecast
Actual
Predicted
Forecast
200
150
100
MAPE:
MAD:
MSD:
100
3.6952
5.6622
52.7851
200
Time
:
Seasonal Indices ( 1)
4 3 2 1 12 11
155
2
5 8
. Box Plot
. Out Liers
)( .
.
) (2

wt = zt Tt , t = 1, 2,..., n
=St + Et , t = 1, 2,..., n
yt = zt St , t = 1, 2,..., n
=Tt + Et , t = 1,2,..., n
Et

et = zt Tt St , t = 1,2,..., n
=Et , t = 1,2,..., n
) (3 24 .
: :
zt = Tt St + Et , t = 1,2,..., n
;MTB > %Decomp 'GasDemand' 12

>SUBC
;Forecasts 24
>SUBC
;"Title "Forecast of Gasoline Demand
SUBC> Start 1.
Executing from file: D:\MTBWIN\MACROS\Decomp.MAC
Macro is running ... please wait

GasDeman
192.000
0
Data
Length
NMissing
Trend Line Equation

Yt = 96.4074 + 0.680579*t
Seasonal Indices
Index
0.860355
0.828555
0.892431
0.936273
1.06124
1.07274
1.15775
156
Period
1
2
3
4
5
6
7
8
9
10
11
12
1.17075
1.03409
1.05059
0.966300
0.968923
Accuracy of Model
MAPE:
MAD:
MSD:
3.6338
5.7720
56.8996
Forecasts
Row
Period
Forecast
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
195.954
189.275
204.474
215.156
244.596
247.977
268.415
272.227
241.154
245.718
226.660
227.935
202.980
196.042
211.762
222.803
253.263
256.738
277.870
281.789
249.599
254.298
234.552
235.848
1.2
Seasonal Indices
Original Data, by Seasonal Period

250
1.1
200
1.0
150
0.9
100
0.8
1 2 3 4 5 6 7 8 9 10 11 12
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 9 10 11 12
Percent Variation, by Seasonal Period
Residuals, by Seasonal Period

20
10
0
157 -10
-20
-30
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 7 8 9 10 11 12
(4)
(5)
Original Data
Detrended Data
250
1.3
1.2
200
1.1
1.0
150
0.9
100
0.8
0
100
200
Seasonally Adjusted Data
240
220
200
180
160
140
120
100
100
200
Seasonally Adj. and Detrended Data

20
10
0
-10
-20
-30
100
200
100
200
(6)
Actual
280
Predicted
GasDeman
Forecast
Actual
Predicted
Forecast
180
MAPE:
MAD:
MSD:
80
0
100
3.6338
5.7720
56.8996
200
Time
:
.(3) ( 2) ( 1) ( 6) ( 5) ( 4)

:
MAPE Mean Absolute Percentage Error -1

158
zt zt
zt
100, zt 0
n
t =1
= MAPE
-2 Mean Absolute Deviation MAD

n
zt
z
t =1
= MAD
-3 ) ( ) MSD ( MSE
2
) zt
( z
t =1
= MSD

MSD MSE .
:
MAPE:
3.6952
MAD:
5.6622
MSD:
52.7851
:
MAPE:
3.6338
MAD:
5.7720
MSD:
56.8996
MSD
.
Decomposition
: Method
.
168
MTB > Read "E:\Mtbwin\milk.dat" c1.
Entering data from file: E:\Mtbwin\milk.dat
168 rows read.
'MTB > name c1='MilkProd
MTB > print c1
Data Display
MilkProd
582
598
634
635
688
698
711
734
751
800
553
565
594
602
645
660
677
690
711
763
577
587
611
621
667
687
706
723
747
800
568
583
604
615
661
681
701
725
740
790
599
617
639
653
697
722
740
764
783
834
640
660
678
702
735
767
783
801
819
869
159
697
716
736
756
798
817
826
845
859
913
727
742
770
782
811
837
858
871
886
942
656
673
705
722
755
784
796
805
824
878
640
653
688
709
736
762
775
785
807
860
561
566
618
622
635
667
696
690
707
756
589
600
628
658
677
713
717
734
750
804
826
821
828
834
799
773
778
782
890
883
889
892
900
898
902
903
961
957
969
966
935
924
947
937
894
881
908
896
855
837
867
858
809
784
815
817
810
791
812
827
766
760
773
797
805
802
813
843
:
MTB > TSPlot 'MilkProd';
SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect.
1000
MilkProd
900
800
700
600
Index
50
100
150
: :
zt = Tt + St + Et , t = 1,2,..., n z1 , z2 ,..., zn
:
:
: Tt -1
Tt zt = a + bt , t = 1, 2,...,168
:
MTB > set c2
DATA> 1:168
DATA> end
MTB > name c1='MilkProd' c2='Time' c3='Trend' c5='Detrend' c6='Index'
c8='Fitted' c9='Resid'
MTB > regr c1 1 c2;
SUBC> fits c3.
Regression Analysis
The regression equation is
MilkProd = 612 + 1.69 Time
Predictor
Constant
Time
S = 60.74
Coef
StDev
T
P
611.682
9.414
64.97
0.000
1.69262
0.09663
17.52
0.000
R-Sq = 64.9%
R-Sq(adj) = 64.7%
Analysis of Variance
Source
Regression
Error
Total
DF
1
166
167
SS
1132003
612439
1744443
MS
1132003
3689
160
F
306.83
P
0.000
900
800
Trend
700
600
150
100
50
Index
-2
Detrended Series zt zt = zt Tt , t = 1,2,...,168
MTB > let c5=c1-c3
100
Detrend
-100
150
100
50
Index
zt Tt = St + Et , t = 1,2,...,168
-3 Seasonal Indices:
I s , s = 1,2,...,12 I1
I 2 d t = zt Tt , t = 1,2,...,168
:
161
1
( d1 + d13 + d 25 + L + d157 )
14
1
I 2 = ( d 2 + d14 + d 26 + L + d158 )
14
M
1
I12 = ( d12 + d 24 + d 36 + L + d168 )
14
I1 =
:
MTB >
DATA>
DATA>
MTB >
SUBC>
SUBC>
MTB >
&
CONT>
MTB >
MTB >
MTB >
DATA>
DATA>
MTB >
set c4
14(1:12)
end
stat c5;
by c4;
mean c6.
Stack 'Index' 'Index' 'Index' 'Index' 'Index' 'Index' 'Index'
'Index' 'Index' 'Index' 'Index' 'Index' 'Index' 'Index' c7.
let c8=c3+c7
let c9=c1-c8
set c10
1:12
end
print c10 c6
Data Display
Row
Season
Index
1
2
3
4
5
6
7
8
9
10
11
12
1
2
3
4
5
6
7
8
9
10
11
12
-18.328
-57.806
34.716
49.595
110.616
82.281
32.517
-9.747
-52.297
-48.775
-79.754
-43.018
z168 ( l) = 612 + 1.69 ( l + 168 ) + I ( l mod 12) , l = 1,2,...
: -4
169
z168 ( 1) = 612 + 1.69 ( 169 ) + I1
=897.61 + ( 18.328 ) = 879.282

:
: Minitab
162
-1
(1) Tt
-2
Detrended Series
-3
-4

: -5
I1 = Median ( d1 , d13 , d 25 ,L, d157 )
I 2 = Median ( d 2 , d14 , d 26 ,L, d158 )
M
I12 = Median ( d12 , d 24 , d 36 ,L, d168 )
-6
:
MTB > Read "E:\Mtbwin\milk.dat" c1.
Entering data from file: E:\Mtbwin\milk.dat
168 rows read.
MTB > name c1='MilkProd'
MTB > set c2
DATA> 1:168
DATA> end
MTB > name c2='Time'
MTB > regr c1 1 c2;
SUBC> fits c3.
Regression Analysis
MilkProd = 612 + 1.69 Time
Predictor
Constant
Time
Coef
611.682
1.69262
S = 60.74
StDev
9.414
0.09663
R-Sq = 64.9%
T
64.97
17.52
P
0.000
0.000
R-Sq(adj) = 64.7%
Analysis of Variance
Source
Regression
Error
Total
DF
1
166
167
SS
1132003
612439
1744443
Unusual Observations
Obs
Time
MilkProd
113
113
942.00
125
125
961.00
MS
1132003
3689
Fit
802.95
823.26
F
306.83
StDev Fit
5.44
6.11
P
0.000
Residual
139.05
137.74
R denotes an observation with a large standardized residual
163
St Resid
2.30R
2.28R
MTB
MTB
MTB
MTB
>
>
>
>
name c3='Trend'
let c4=c1-c3
name c4='Detrend'
Name c5 = 'AVER1'
: 12
MTB > %MA 'Detrend' 12;
SUBC> Center;
SUBC>
Averages 'AVER1'.
Moving average
Data
Length
NMissing
Detrend
168.000
0
Moving Average
Length: 12
Accuracy Measures
MAPE: 111.68
MAD:
52.36
MSD: 3564.77

MTB >
MTB >
MTB >
DATA>
DATA>
MTB >
SUBC>
SUBC>
MTB >
let c6=c4-c5
name c6='DeSeason'
set c2
14(1:12)
end
stat c6;
by c2;
median c7.
name c7='SeasInx'
Data Display
Row
Season
SeasInx
1
2
3
4
5
6
7
8
9
10
11
12
1
2
3
4
5
6
7
8
9
10
11
12
-20.750
-58.958
35.625
50.083
109.542
81.292
33.917
-10.000
-52.792
-50.250
-79.958
-44.375
:
MTB > %Decomp 'MilkProd' 12;
164
SUBC> Additive ;
SUBC> Start 1.
Executing from file: E:\MTBWIN\MACROS\Decomp.MAC

Data
Length
NMissing
MilkProd
168.000
0
Trend Line Equation

Yt = 611.682 + 1.69262*t
Seasonal Indices
Period
Index
1
2
3
4
5
6
7
8
9
10
11
12
-20.1979
-58.4062
36.1771
50.6354
110.094
81.8437
34.4687
-9.44792
-52.2396
-49.6979
-79.4063
-43.8229
Accuracy of Model
MAPE:
MAD:
MSD:
1.583
12.088
244.406
.
: Decomp%
MTB > %Decomp 'MilkProd' 12;
SUBC> Additive ;
SUBC>
Forecasts 12;
SUBC> Start 1.
:
Forecasts
Row
Period
Forecast
1
2
3
4
5
6
169
170
171
172
173
174
877.54
841.02
937.30
953.45
1014.60
988.04
165
942.36
900.13
859.04
863.27
835.25
872.53
175
176
177
178
179
180
7
8
9
10
11
12
166

Using Moving
Average Smoothing for Forecasting

z1 , z2 , z3 ,K , zn2 , zn1 , zn m

1
zt = ( zt + zt 1 + zt 2 + L + zt m+1 ) , t = m, m + 1,..., n
m
1
zt = zt 1 + ( zt zt m ) , t = m, m + 1,..., n
m
. n m + 1
m=3
1
) ( z3 + z2 + z1
3
1
1
) z4 = ( z4 + z3 + z2 ) or z4 = z3 + ( z4 z1
3
3
M
1
1
) zn = ( zn + zn 1 + zn 2 ) or zn = zn 1 + ( zn zn 3
3
3
= z3
zt = + at , at : WN ( 0, 2 ) , t = 1,2,..., n
V ( z t ) = , t
2
, t = m, m + 1,..., n
m
m
.
: m .
2
= ) V ( zt
:
:
zn ( l) = zn1 , l > 0
:
MINITAB EMPLOY.MTW
'E:\Mtbwin\DATA\EMPLOY.MTW'.
MTB > Retrieve

MTB > info
167
Information on the Worksheet

Column
C1
C2
C3
Count
60
60
60
Name
Trade
Food
Metals
Metals
Metals
44.2
43.1
42.4
42.9
45.0
49.4
50.5
44.3
42.6
43.1
43.6
45.5
50.0
51.2
44.4
42.4
43.2
44.7
46.2
50.0
50.7
43.4
42.2
42.8
44.5
46.8
49.6
50.3
42.8
41.8
43.0
45.0
47.5
49.9
49.2
44.3
40.1
42.8
44.8
48.3
49.6
48.1
44.4
42.0
42.5
44.9
48.3
50.7
44.8
42.4
42.6
45.2
49.1
50.7
44.4
43.1
42.3
45.2
48.9
50.9
:
MTB > TSPlot 'Metals';
SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect.
Metals
50
45
40
Index
10
20
30
40
50
60
m=3
: 6
MTB > %MA 'Metals' 3;
SUBC>
Forecasts 6;
SUBC>
Title "Smoothing and Forecasting Metals".
Moving average
Data
Length
NMissing
Metals
60.0000
0
168
Moving Average
Length: 3
Accuracy Measures
MAPE: 1.55036
MAD: 0.70292
MSD: 0.76433
Row
1
2
3
4
5
6
Period
Forecast
61
62
63
64
65
66
49.2
49.2
49.2
49.2
49.2
49.2
Lower
47.4865
47.4865
47.4865
47.4865
47.4865
47.4865
Upper
50.9135
50.9135
50.9135
50.9135
50.9135
50.9135
Smoothing and Forecasting Metals
Actual
Predicted
Metals
50
Forecast
Actual
Predicted
Forecast
45
Moving Average
Length:
MAPE: 1.55036
40
0
10
20
30
40
50
MAD:
0.70292
MSD:
0.76433
60
Time
50.3 + 49.2 + 48.1 147.6

=
= 49.2
3
3
zn +1 , zn+2 ,..., zn+6 6
: z61 , z62 ,..., z66
z60 ( 1) = z60 ( 2 ) = L = z60 ( 6 ) = 49.2
z59 =
zn ( l) 1.96 , l > 0 95%

MSD = 0.76433 [ 49.2 1.96 ]
95% = 0.8743 2 = 0.76433 2
:
49.2 1.96 ( 0.8743) = [ 49.2 1.7135] = [ 47.4865,50.9135]
169
z60+l [ 47.4865,50.9135] , l > 0 with probability 0.95
: MSD
) zi
n 1
( z
n2
i =2
= MSD =
2
:
5 7
.
Running Median
Outliers
m
18
13
15
11
12
1500
10
)z(t
5
7
20

1500
1000
)z(t
500
0
10
Index
3
Moving Average
Actual
1500
Predicted
Actual
Predicted
1000
)z(t
Moving Average
3
1081
500
Length:
MAPE:
273
MAD:
268811
MSD:
0
15
10
Time
170

.
j = 2i + 1 z1 , z2 , z3 ,K , zn2 , zn1 , zn

z%t = med ( zt i ,..., zt ,..., zt +i ) , j = 2i + 1
j = 3 ) z%t = med ( zt 1 , zt , zt +1 3

15
)smoothz(t
10
5
12
10
Index
z9 15 1500
20
15
)z(t
10
10
171
Index

Using Single
: Exponential Smoothing for Forecasting

.
z1 , z2 , z3 ,K , zn2 , zn1 , zn m

1
zt = ( zt + zt 1 + zt 2 + L + zt m+1 ) , t = m, m + 1,..., n
m

1
1
1
1
zt = zt + zt 1 + zt 2 + L + zt m+1 , t = m, m + 1,..., n
m
m
m
m
1
= zt = zt + zt 1 + zt 2 + L + zt m+1 , t = m, m + 1,..., n,
m

zn
st = zt + ( 1 ) zt 1 + ( 1 ) zt 2 + L , t = 1, 2,..., n, 0 < < 1
2
st

st = zt + ( 1 ) st 1 , t = 1, 2,..., n, s0 = z
zn ( l) = sn , l 1
EMPLOY.MTW
MTB > Retrieve
Metals
44.4
43.1
42.3
45.2
48.9
50.9
44.8
42.4
42.6
45.2
49.1
50.7
44.4
42.0
42.5
44.9
48.3
50.7
44.3
40.1
42.8
44.8
48.3
49.6
48.1
42.8
41.8
43.0
45.0
47.5
49.9
49.2
43.4
42.2
42.8
44.5
46.8
49.6
50.3
44.4
42.4
43.2
44.7
46.2
50.0
50.7
44.3
42.6
43.1
43.6
45.5
50.0
51.2
Metals
44.2
43.1
42.4
42.9
45.0
49.4
50.5
;'MTB > TSPlot 'Metals

>SUBC
;Index
>SUBC
;TDisplay 11
>SUBC
;Symbol
172
SUBC>
Connect.
M
etals
50
45
40
Index
10
20
30
40
50
60
= 0.2
: 6
MTB > %SES 'Metals';
SUBC>
Weight 0.2;
SUBC>
Forecasts 6;
SUBC>
Title "Smoothing and Forecasting Metals";
SUBC>
Initial 6.
Single Exponential Smoothing
Data
Metals
Length
60.0000
NMissing
0
Smoothing Constant
Alpha: 0.2
Accuracy Measures
MAPE: 2.17304
MAD: 1.00189
MSD: 1.45392
Row
Period
Forecast
Lower
Upper
1
2
3
4
5
6
61
62
63
64
65
66
49.7216
49.7216
49.7216
49.7216
49.7216
49.7216
47.2670
47.2670
47.2670
47.2670
47.2670
47.2670
52.1763
52.1763
52.1763
52.1763
52.1763
52.1763
Smoothing and Forecasting Metals
Actual
Predicted
Forecast
Metals
50
Actual
Predicted
Forecast
45
Smoothing Constant
Alpha:
40
0
10
20
30
40
Time
173
50
60
0.200
MAPE:
2.17304
MAD:
1.00189
MSD:
1.45392
:
-1
:
z1 , z2 ,K , zn 1 , zn 2 = 0.2
si = zi + ( 1 ) si 1 , i = 1,2,..., n
s0
s0
m
) , m = 6 ( or n, if n<6
i =1
= s0
44.2 + 44.3 + 44.4 + 43.4 + 42.8 + 44.3

= 43.9
6
= s0

s1 = z1 + ( 1 ) s0 = 0.2 ( 44.2 ) + 0.8 ( 43.9 ) = 8.84 + 35.12 = 43.96
s2 = z2 + ( 1 ) s1 = 0.2 ( 44.3) + 0.8 ( 43.96 ) = 8.86 + 35.168 = 44.028
:
RESI1
FITS1
SMOO1
Metals
Time
0.30000
0.34000
0.37200
-0.70240
-1.16192
0.57046
0.55637
0.84510
0.27608
-1.07914
-1.36331
-1.29065
-1.23252
-1.38601
-2.80881
-0.34705
0.12236
0.79789
-0.06169
0.65065
0.62052
0.09642
0.27713
0.02171
-0.28264
-0.12611
-0.40089
0.27929
0.92343
1.83875
1.27100
1.51680
1.01344
0.91075
1.02860
43.9000
43.9600
44.0280
44.1024
43.9619
43.7295
43.8436
43.9549
44.1239
44.1791
43.9633
43.6906
43.4325
43.1860
42.9088
42.3470
42.2776
42.3021
42.4617
42.4494
42.5795
42.7036
42.7229
42.7783
42.7826
42.7261
42.7009
42.6207
42.6766
42.8613
43.2290
43.4832
43.7866
43.9892
44.1714
43.9600
44.0280
44.1024
43.9619
43.7295
43.8436
43.9549
44.1239
44.1791
43.9633
43.6906
43.4325
43.1860
42.9088
42.3470
42.2776
42.3021
42.4617
42.4494
42.5795
42.7036
42.7229
42.7783
42.7826
42.7261
42.7009
42.6207
42.6766
42.8613
43.2290
43.4832
43.7866
43.9892
44.1714
44.3771
44.2
44.3
44.4
43.4
42.8
44.3
44.4
44.8
44.4
43.1
42.6
42.4
42.2
41.8
40.1
42.0
42.4
43.1
42.4
43.1
43.2
42.8
43.0
42.8
42.5
42.6
42.3
42.9
43.6
44.7
44.5
45.0
44.8
44.9
45.2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
174
0.82288
0.45830
0.86664
1.39331
1.71465
2.07172
2.45738
1.96590
2.37272
1.69818
1.85854
2.08683
1.66947
0.93557
1.04846
0.53877
1.53101
1.22481
1.17985
0.54388
1.13510
0.40808
-0.07353
-1.15883
-2.02706
44.3771
44.5417
44.6334
44.8067
45.0853
45.4283
45.8426
46.3341
46.7273
47.2018
47.5415
47.9132
48.3305
48.6644
48.8515
49.0612
49.1690
49.4752
49.7202
49.9561
50.0649
50.2919
50.3735
50.3588
50.1271
44.5417
44.6334
44.8067
45.0853
45.4283
45.8426
46.3341
46.7273
47.2018
47.5415
47.9132
48.3305
48.6644
48.8515
49.0612
49.1690
49.4752
49.7202
49.9561
50.0649
50.2919
50.3735
50.3588
50.1271
49.7216
45.2
45.0
45.5
46.2
46.8
47.5
48.3
48.3
49.1
48.9
49.4
50.0
50.0
49.6
49.9
49.6
50.7
50.7
50.9
50.5
51.2
50.7
50.3
49.2
48.1
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
SMOO1 si , i = 1,2,...,60 FITS1

zi = si 1 , i = 1,2,...,60 RESI1
) (Residuals ei = zi zi , i = 1,2,...,60
-2 :
zn ( l) = sn , l > 0

z60 ( l) = 49.7216, l > 0
6 zn +1 , zn+2 ,..., zn+6
z61 , z62 ,..., z66:
z60 ( 1) = z60 ( 2 ) = L = z60 ( 6 ) = 49.7216
-3 95% zn ( l) 1.96 , l > 0
] [ 49.7216 1.96 MSD = 1.45392
2 2 = 1.45392 = 1.205786 95%
:
] 49.7216 1.96 ( 1.205786 ) = [ 49.7216 2.3633] = [ 47.35826,52.08494
z60+l [ 47.3582,50.0849] , l > 0 with probability 0.95

: MSD
175
) zi
( z
n 1
i =1
= MSD =
2
:
= 0.3,0.4,0.5
.
176

Using Double
: Exponential Smoothing for Forecasting
: : Browns Method
z1 , z2 ,K , zn 1 , zn 2 0 < < 1 :
1
= zt + ( 1 ) st( 1) , t = 1,2,..., n
)( 1
st
1
1
) ( st ) st( ) = st ( ) ( 1
)( 2
)( 1
)( 2
st = st + ( 1 ) st 1 , t = 1, 2,..., n
2
) ( st ) ( 2
at = 2 st( ) st( ) , t = 1,2,..., n
1
2
= bt
st( ) st( ) , t = 1, 2,..., n
1
2
t = 1,2,..., n
zn +l , l > 0
zt = at + bt t ,
zn ( l) = an + bn l, l > 0
) ( s0 ) ( : s0
1
1
b0
1
2
s0( ) = a0 2
b0
zt = + t + et ,
s0( ) = a0
1
a0 b0 t = 1,2,..., n
a0 = b0 =
: : Holts Method
z1 , z2 ,K , zn1 , zn2 0 < < 1 0 < < 1 :

st = zt + ( 1 ) ( st 1 + bt 1 ) , t = 1,2,..., n
t = 1, 2,..., n

bt = ( st st 1 ) + ( 1 ) bt 1 ,
t = 1,2,..., n
zt = st + bt t ,
zn ( l) = sn + bn l, l > 0
s0 b0
177
s0 = z1
b0 = z2 z1
b0 =
or
( z2 z1 ) + ( z3 z2 ) ( z3 z1 )
=
or
2
2
( z z ) + ( z3 z2 ) + ( z4 z3 ) ( z4 z1 )
b0 = 2 1
=
3
3
:
EMPLOY.MTW
MTB > Retrieve 'E:\Mtbwin\DATA\EMPLOY.MTW'.
Metals
44.2
43.1
42.4
42.9
45.0
49.4
50.5
Metals
44.3
42.6
43.1
43.6
45.5
50.0
51.2
44.4
42.4
43.2
44.7
46.2
50.0
50.7
43.4
42.2
42.8
44.5
46.8
49.6
50.3
42.8
41.8
43.0
45.0
47.5
49.9
49.2
44.3
40.1
42.8
44.8
48.3
49.6
48.1
44.4
42.0
42.5
44.9
48.3
50.7
50
Metals
44.4
43.1
42.3
45.2
48.9
50.9
MTB > TSPlot 'Metals';

SUBC>
Index;
SUBC>
TDisplay 11;
SUBC>
Symbol;
SUBC>
Connect.
45
40
Index
44.8
42.4
42.6
45.2
49.1
50.7
10
20
30
178
40
50
60

WEIGHT Macro) %DES)
0.2
MTB > %DES 'Metals';
SUBC>
Weight 0.2 0.2;
SUBC>
Forecasts 6;
SUBC>
Title "Brown's Double Exponential Smoothing";
SUBC>
Table.
Double Exponential Smoothing

Data
Length
NMissing
Metals
60.0000
0
Smoothing Constants
Alpha (level): 0.2
Gamma (trend): 0.2
Accuracy Measures
MAPE: 2.16187
MAD: 0.97032
MSD: 1.62936
Time Metals
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
44.2
44.3
44.4
43.4
42.8
44.3
44.4
44.8
44.4
43.1
42.6
42.4
42.2
41.8
40.1
42.0
42.4
43.1
42.4
43.1
43.2
42.8
43.0
42.8
42.5
42.6
42.3
42.9
43.6
44.7
Smooth
Predict
Error
41.7739
42.4976
43.1686
43.5546
43.7373
44.1459
44.4990
44.8575
45.0620
44.9391
44.6673
44.3271
43.9378
43.4769
42.7011
42.3565
42.1465
42.1286
42.0132
42.0763
42.1877
42.2374
42.3396
42.4078
42.4181
42.4495
42.4207
42.5129
42.7420
43.1797
41.1674
42.0470
42.8607
43.5933
43.9716
44.1074
44.5238
44.8719
45.2275
45.3989
45.1841
44.8088
44.3723
43.8962
43.3514
42.4456
42.0831
41.8857
41.9164
41.8204
41.9347
42.0967
42.1745
42.3098
42.3976
42.4119
42.4509
42.4161
42.5276
42.7996
3.03257
2.25303
1.53927
-0.19330
-1.17163
0.19257
-0.12377
-0.07189
-0.82751
-2.29891
-2.58407
-2.40884
-2.17229
-2.09617
-3.25142
-0.44557
0.31694
1.21426
0.48355
1.27964
1.26533
0.70327
0.82549
0.49024
0.10244
0.18809
-0.15090
0.48394
1.07245
1.90036
179
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
44.5
45.0
44.8
44.9
45.2
45.2
45.0
45.5
46.2
46.8
47.5
48.3
48.3
49.1
48.9
49.4
50.0
50.0
49.6
49.9
49.6
50.7
50.7
50.9
50.5
51.2
50.7
50.3
49.2
48.1
43.5507
43.9854
44.3338
44.6511
44.9749
45.2430
45.4157
45.6373
45.9491
46.3285
46.7909
47.3492
47.8339
48.4002
48.8413
49.2966
49.7849
50.1841
50.4162
50.6292
50.7104
50.9509
51.1334
51.3019
51.3407
51.4782
51.4770
51.3649
51.0127
50.4384
43.3133
43.7317
44.2172
44.5889
44.9187
45.2538
45.5197
45.6716
45.8863
46.2106
46.6136
47.1115
47.7173
48.2253
48.8267
49.2707
49.7311
50.2302
50.6202
50.8114
50.9880
51.0137
51.2417
51.4024
51.5509
51.5477
51.6712
51.6311
51.4659
51.0230
1.18668
1.26826
0.58280
0.31112
0.28133
-0.05376
-0.51967
-0.17162
0.31368
0.58938
0.88637
1.18850
0.58266
0.87469
0.07332
0.12930
0.26890
-0.23017
-1.02022
-0.91145
-1.38797
-0.31368
-0.54169
-0.50244
-1.05093
-0.34770
-0.97120
-1.33115
-2.26587
-2.92300
Row
Period
Forecast
Lower
Upper
1
2
3
4
5
6
61
62
63
64
65
66
50.3318
50.2252
50.1186
50.0120
49.9054
49.7988
47.9545
47.7984
47.6384
47.4749
47.3080
47.1381
52.7091
52.6520
52.5987
52.5490
52.5027
52.4594
Brown's Double Exponential Smoothing
Actual
Predicted
Forecast
Metals
50
Actual
Predicted
Forecast
45
Smoothing Constants
Alpha (level): 0.200
Gamma (trend):0.200
MAPE:
MAD:
MSD:
40
0
10
20
30
Time
180
40
50
60
2.16187
0.97032
1.62936
: b0 a0
MTB >
DATA>
DATA>
MTB >
set c4
1:60
end
regr c3 1 c4
Regression Analysis
Metals = 41.0 + 0.152 C4
b0 = 0.152 a0 = 41.0
1
0.8
b0 = 41.0
( 0.152 ) = 41.608
0.2
1
0.8
= a0 2
b0 = 41.0 2
( 0.152 ) = 42.216
0.2
s0( ) = a0
1
s0(
2)
s1( ) = ( 0.2 ) ( 44.2 ) + ( 0.8 ) ( 41.608 ) = 42.1264

1
s1( ) = ( 0.2 ) ( 42.1264 ) + ( 0.8 ) ( 42.216 ) = 42.19808

2
a1 = ( 2 ) ( 42.1264 ) 42.19808 = 42.05472
( 0.2 ) 42.19808 42.05472 = 0.03584

(
)
( 0.8)
z1 = 42.05472 + ( 0.03584 ) ( 1) = 42.09056
b1 =

. MSD
:
WEIGHT Macro) %DES)
= 0.3 = 0.2
MTB > RETR 'E:\Mtbwin\DATA\EMPLOY.MTW'.

Retrieving worksheet from file: E:\Mtbwin\DATA\EMPLOY.MTW
Worksheet was saved on 6/ 5/1996
MTB > %DES 'Metals';
SUBC>
Weight 0.2 0.3;
SUBC>
Forecasts 6;
SUBC>
Title "Holt's Double Exponential Smoothing";
SUBC>
Table.
Double Exponential Smoothing

Data
Length
NMissing
Metals
60.0000
0
181
Smoothing Constants
Alpha (level): 0.2
Gamma (trend): 0.3
Accuracy Measures
MAPE: 2.15656
MAD: 0.96328
MSD: 1.56274
Time
Metals
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
44.2
44.3
44.4
43.4
42.8
44.3
44.4
44.8
44.4
43.1
42.6
42.4
42.2
41.8
40.1
42.0
42.4
43.1
42.4
43.1
43.2
42.8
43.0
42.8
42.5
42.6
42.3
42.9
43.6
44.7
44.5
45.0
44.8
44.9
45.2
45.2
45.0
45.5
46.2
46.8
47.5
48.3
48.3
49.1
48.9
49.4
50.0
50.0
Smooth
41.7739
42.5461
43.2891
43.7501
43.9779
44.3895
44.7334
45.0685
45.2405
45.0676
44.7113
44.2595
43.7466
43.1634
42.2751
41.8139
41.5361
41.5057
41.4371
41.5799
41.8054
41.9895
42.2254
42.4205
42.5395
42.6522
42.6793
42.7982
43.0394
43.4861
43.8762
44.3257
44.6858
45.0007
45.3066
45.5449
45.6749
45.8384
46.0888
46.4159
46.8406
47.3799
47.8665
48.4419
48.9016
49.3693
49.8653
50.2702
Predict
41.1674
42.1076
43.0113
43.8376
44.2724
44.4118
44.8167
45.1356
45.4506
45.5595
45.2391
44.7244
44.1332
43.5043
42.8188
41.7674
41.3202
41.1072
41.1964
41.1999
41.4568
41.7869
42.0317
42.3257
42.5493
42.6653
42.7741
42.7728
42.8993
43.1826
43.7202
44.1571
44.6573
45.0259
45.3333
45.6312
45.8436
45.9230
46.0610
46.3199
46.6757
47.1499
47.7582
48.2774
48.9020
49.3617
49.8317
50.3378
Error
3.03257
2.19238
1.38868
-0.43760
-1.47237
-0.11184
-0.41671
-0.33560
-1.05057
-2.45952
-2.63911
-2.32443
-1.93322
-1.70426
-2.71884
0.23263
1.07985
1.99283
1.20365
1.90008
1.74322
1.01314
0.96829
0.47431
-0.04933
-0.06528
-0.47413
0.12724
0.70070
1.51743
0.77977
0.84285
0.14275
-0.12590
-0.13327
-0.43116
-0.84361
-0.42295
0.13895
0.48014
0.82428
1.15014
0.54181
0.82265
-0.00205
0.03832
0.16832
-0.33779
182
49
50
51
52
53
54
55
56
57
58
59
60
49.6
49.9
49.6
50.7
50.7
50.9
50.5
51.2
50.7
50.3
49.2
48.1
50.4979
50.6862
50.7296
50.9166
51.0532
51.1813
51.1869
51.2901
51.2673
51.1350
50.7591
50.1448
50.7224
50.8827
51.0121
50.9708
51.1415
51.2516
51.3586
51.3127
51.4092
51.3438
51.1489
50.6560
-1.12240
-0.98275
-1.41206
-0.27079
-0.44152
-0.35162
-0.85860
-0.11267
-0.70916
-1.04381
-1.94889
-2.55603
Row
Period
Forecast
Lower
Upper
1
2
3
4
5
6
61
62
63
64
65
66
49.8884
49.6319
49.3755
49.1190
48.8626
48.6061
47.5283
47.1597
46.7803
46.3915
45.9946
45.5908
52.2484
52.1041
51.9707
51.8466
51.7306
51.6215
Holt's Double Exponential Smoothing
Actual
Predicted
Forecast
Metals
50
Actual
Predicted
Forecast
45
Smoothing Constants
Gamma (trend):0.300
MAPE:
MAD:
MSD:
40
0
10
20
30
40
50
2.15656
0.96328
1.56274
60
Time
: ) :

(
183

Triple Exponential Smoothing: Winters' ThreeParameter Trend and Seasonality Smoothing Method

Decomposition Method Winters' trend and
seasonal smoothing

:
zt
+ ( 1 ) ( st 1 + bt 1 ) , t = 1,2,..., n
St s
:
st =
bt = ( st st 1 ) + ( 1 ) bt 1 , t = 1,2,..., n
:
zt
+ ( 1 ) St s , t = 1, 2,..., n
st
Si i s

zt = ( st + bt t ) St s , t = 1, 2,..., n

zn ( l) = ( sn + bn l) Sn s +l , l > 0
St =

.
:
EMPLOY.MTW
MTB > Retrieve
Food
52.1
51.5
53.5
54.3
53.4
52.3
53.1
54.6
55.3
53.6
53.3
54.2
58.2
55.3
53.9
54.8
66.9
58.5
55.6
55.8
Food
70.7
69.3
60.1
57.9
57.7
65.3
69.6
68.9
62.6
60.5
184
56.5
64.2
68.8
70.3
66.4
53.4
55.5
63.6
69.4
75.5
52.5
53.3
57.1
64.7
73.3
53.2
52.4
52.2
57.1
68.1
53.0
51.5
51.5
53.9
58.1
53.5
51.5
51.7
53.5
54.8
75
Food
70
65
60
55
50
Index
10
20
30
40
50
60
: 12
Additive Model :
zt = bt + St + et , t = 1,2,..., n
MTB > %Wintadd 'Food' 12;
SUBC>
Weight 0.2 0.2 0.2;
SUBC>
Forecasts 12;
SUBC>
Title "Wintrs' Trend and Seasonal Smoothing";
SUBC>
Table.
Winters' additive model

Data
Length
NMissing
Food
60.0000
0
Smoothing Constants
Alpha (level):
0.2
Gamma (trend):
0.2
Delta (seasonal): 0.2
Accuracy Measures
MAPE: 1.94769
MAD: 1.15100
MSD: 2.66711
185
Time
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Food
53.5
53.0
53.2
52.5
53.4
56.5
65.3
70.7
66.9
58.2
55.3
53.4
52.1
51.5
51.5
52.4
53.3
55.5
64.2
69.6
69.3
58.5
55.3
53.6
52.3
51.5
51.7
51.5
52.2
57.1
63.6
68.8
68.9
60.1
55.6
53.9
53.3
53.1
53.5
53.5
53.9
57.1
64.7
69.4
70.3
62.6
57.9
55.8
54.8
54.2
54.6
54.3
54.8
58.1
68.1
73.3
75.5
66.4
60.5
Smooth
48.7755
49.6020
51.0736
52.0733
53.5117
57.4851
66.2299
71.7852
71.8932
62.3206
57.5208
55.1544
55.0393
53.6493
53.1185
52.2661
52.6528
55.7616
63.8483
68.8787
68.0386
59.2825
55.0979
53.0432
53.0377
52.1394
51.9889
51.7214
52.1875
55.0750
63.7515
68.8427
68.0160
58.9061
55.3220
53.4419
53.3084
52.6717
52.9095
52.9745
53.7952
57.2065
65.2747
70.4039
69.6200
60.5655
57.0392
55.3721
55.2403
54.6681
54.8138
54.7366
55.3001
58.5310
66.4168
71.8806
71.8794
63.7240
60.3141
Predict
49.4303
50.4197
51.9944
53.0424
54.4591
58.3901
67.0593
72.5443
72.5785
62.7787
57.7958
55.3296
55.1373
53.6258
53.0100
52.0971
52.4960
55.6369
63.7181
68.7678
67.9610
59.2585
55.0435
52.9991
53.0177
52.0907
51.9165
51.6403
52.1009
54.9923
63.7531
68.8382
68.0099
58.9356
55.3981
53.5262
53.4076
52.7666
53.0177
53.1020
53.9386
57.3484
65.4066
70.5076
69.6794
60.6497
57.2014
55.5623
55.4399
54.8422
54.9622
54.8705
55.4112
58.6176
66.4827
72.0112
72.0616
64.0437
60.7281
Error
4.06965
2.58027
1.20556
-0.54244
-1.05914
-1.89013
-1.75932
-1.84430
-5.67851
-4.57874
-2.49577
-1.92957
-3.03734
-2.12584
-1.50996
0.30287
0.80401
-0.13695
0.48187
0.83218
1.33902
-0.75851
0.25650
0.60092
-0.71765
-0.59067
-0.21651
-0.14031
0.09913
2.10774
-0.15314
-0.03822
0.89007
1.16436
0.20190
0.37384
-0.10757
0.33345
0.48233
0.39803
-0.03858
-0.24838
-0.70661
-1.10758
0.62065
1.95031
0.69858
0.23773
-0.63993
-0.64220
-0.36218
-0.57048
-0.61119
-0.51765
1.61731
1.28878
3.43843
2.35629
-0.22810
186
60
57.7
58.6397
59.0446
-1.34455
Row
Period
Forecast
Lower
Upper
1
2
3
4
5
6
7
8
9
10
11
12
61
62
63
64
65
66
67
68
69
70
71
72
58.6167
58.3236
58.8195
58.9840
59.8723
63.4804
72.0757
77.4486
77.7540
68.9067
64.6434
62.7731
55.7968
55.4449
55.8775
55.9746
56.7913
60.3243
68.8410
74.1321
74.3528
65.4180
61.0647
59.1020
61.4366
61.2023
61.7614
61.9935
62.9532
66.6365
75.3104
80.7651
81.1552
72.3954
68.2221
66.4441
Wintrs' Trend and Seasonal Smoothing
Actual
80
Predicted
Forecast
Actual
Predicted
Forecast
Food
70
60
Smoothing Constants
Gamma (trend):0.200
Delta (season):0.200
50
MAPE:
MAD:
MSD:
10
20
30
40
50
60
1.94769
1.15100
2.66711
70
Time
zt = bt St + et , t = 1,2,..., n
Multiplicative Model :
MTB > %Wintmult 'Food' 12;

SUBC>
Weight 0.2 0.2 0.2;
SUBC>
Forecasts 12;
SUBC>
Title "Winters' Trend and Seasonal Smoothing";
SUBC>
Table.
Winters' multiplicative model

Data
Food
187
Length
NMissing
60.0000
0
Smoothing Constants
Alpha (level):
0.2
Gamma (trend):
0.2
Delta (seasonal): 0.2
Accuracy Measures
MAPE: 1.88377
MAD: 1.12068
MSD: 2.86696
Time
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Food
53.5
53.0
53.2
52.5
53.4
56.5
65.3
70.7
66.9
58.2
55.3
53.4
52.1
51.5
51.5
52.4
53.3
55.5
64.2
69.6
69.3
58.5
55.3
53.6
52.3
51.5
51.7
51.5
52.2
57.1
63.6
68.8
68.9
60.1
55.6
53.9
53.3
53.1
53.5
53.5
53.9
57.1
64.7
69.4
Smooth
48.7870
49.6755
51.1521
52.1675
53.6181
57.6509
66.6199
72.4105
72.5679
62.7837
57.9154
55.5108
54.4920
53.2117
52.8118
52.0929
52.5894
55.7388
63.7189
68.7087
67.9722
59.4594
55.4037
53.4103
52.6818
51.8659
51.8002
51.6271
52.1643
55.0424
63.6079
68.6702
67.9561
59.1021
55.6210
53.7881
53.0479
52.4502
52.7444
52.8747
53.7689
57.2790
65.4702
70.6713
Predict
49.3853
50.4303
52.0132
53.0746
54.5132
58.5541
67.5607
73.3280
73.3777
63.2634
58.1732
55.6485
54.5392
53.1621
52.6957
51.9302
52.4439
55.6209
63.5782
68.5838
67.8890
59.4361
55.3468
53.3536
52.6356
51.8071
51.7290
51.5549
52.0890
54.9676
63.6199
68.6823
67.9727
59.1487
55.7003
53.8609
53.1211
52.5294
52.8467
53.0029
53.9188
57.4374
65.6357
70.8095
Error
4.11470
2.56966
1.18677
-0.57458
-1.11323
-2.05414
-2.26072
-2.62800
-6.47768
-5.06337
-2.87320
-2.24849
-2.43920
-1.66212
-1.19573
0.46985
0.85611
-0.12087
0.62178
1.01617
1.41104
-0.93606
-0.04680
0.24639
-0.33562
-0.30705
-0.02902
-0.05492
0.11103
2.13244
-0.01988
0.11774
0.92727
0.95133
-0.10032
0.03912
0.17892
0.57055
0.65329
0.49714
-0.01879
-0.33743
-0.93567
-1.40954
188
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
70.3
62.6
57.9
55.8
54.8
54.2
54.6
54.3
54.8
58.1
68.1
73.3
75.5
66.4
60.5
57.7
69.8908
60.7552
57.1925
55.5181
54.8764
54.3244
54.5372
54.5298
55.1925
58.6054
66.7739
72.4056
72.3385
63.6729
60.0395
58.3023
69.9719
60.8370
57.3348
55.6775
55.0383
54.4749
54.6769
54.6661
55.3155
58.7141
66.8698
72.5622
72.5236
63.9378
60.3781
58.6338
0.32815
1.76302
0.56523
0.12253
-0.23826
-0.27486
-0.07694
-0.36612
-0.51551
-0.61410
1.23016
0.73784
2.97638
2.46217
0.12191
-0.93381
Row
Period
Forecast
Lower
Upper
1
2
3
4
5
6
7
8
9
10
11
12
61
62
63
64
65
66
67
68
69
70
71
72
57.8102
57.3892
57.8332
57.9307
58.8311
62.7415
72.1849
78.1507
78.5092
68.6689
63.9258
61.8189
55.0645
54.5864
54.9687
55.0005
55.8313
59.6686
69.0354
74.9215
75.1976
65.2721
60.4414
58.2446
60.5558
60.1921
60.6977
60.8609
61.8309
65.8145
75.3344
81.3798
81.8208
72.0657
67.4103
65.3933
Winters' Trend and Seasonal Smoothing
Actual
80
Predicted
Forecast
Actual
Predicted
Forecast
Food
70
60
Smoothing Constants
Gamma (trend):0.200
Delta (season):0.200
50
MAPE:
MAD:
MSD:
10
20
30
40
50
60
1.88377
1.12068
2.86696
70
Time

Optimization
189
) (
.
: . = = = 0.2
MSD
:
CPI.MTW MINITAB
'C:\MTBWIN\STUDENT9\CPI.MTW
.'MTB > Retrieve
CPIChange
2.9
6.2
13.5
3.6
1.6
3.2
11.3
1.9
1.3
5.7
6.5
4.3
4.2
1.3
4.4
7.6
3.6
3.0
1.0
4.2
9.1
6.2
4.8
1.0
5.5
5.8
3.2
5.4
CPIChnge
1.7
3.1
11.0
10.3
4.1

MTB > TSPlot
14
12
10
CPIChnge
4
2
0
30
20
25
10
15
Index
MTB > %acf c2

Autocorrelation Function for CPIChnge
Corr
Lag
LBQ
41.23
8 -0.16 -0.51
190
2
Corr
1
Lag
LBQ
22.42
30.32
33.59
36.24
38.97
40.05
40.07
1 0.79 4.53
2 0.46 1.77
3 0.29 1.02
4 0.26 0.88
5 0.26 0.86
6 0.16 0.52
7 -0.02 -0.06
Autocorrelation
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
MTB > %pacf c2

Partial Autocorrelation Function for CPIChnge

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
3
Lag PAC
1
2
3
4
5
6
7
0.79
-0.42
0.35
-0.04
0.08
-0.28
-0.05
Lag PAC
4.53
-2.44
2.01
-0.24
0.43
-1.62
-0.29
8 -0.09
-0.52

5 ( ARMA(1,1

MTB >
SUBC>
SUBC>
SUBC>
SUBC>
SUBC>
SUBC>
arima 1 0 1 c2;
fore 5 c3 c4 c5;
gser;
gacf;
gpacf;
ghist;
gnormal.
ARIMA Model
ARIMA model for CPIChnge
Iteration
SSE
Parameters
0
323.251
0.100
0.100
1
200.616
0.250
-0.050
2
182.146
0.184
-0.200
3
163.067
0.135
-0.350
4
142.864
0.107
-0.500
5
121.402
0.111
-0.650
6
99.668
0.150
-0.800
7
77.036
0.268
-0.950
8
67.550
0.418
-0.956
9
62.802
0.568
-0.964
10
62.108
0.637
-0.973
191
4.522
3.745
4.067
4.308
4.434
4.407
4.197
3.590
2.828
2.062
1.687
11
62.030
0.644
-0.979
1.619
12
62.003
0.647
-0.982
1.584
13
61.996
0.651
-0.985
1.549
14
61.996
0.651
-0.986
1.539
Unable to reduce sum of squares any further
Type
Coef
StDev
AR
1
0.6513
0.1434
MA
1
-0.9857
0.0516
Constant
1.5385
0.4894
Mean
4.412
1.403
T
4.54
-19.11
3.14
Residuals:
MS = 2.0613 DF = 30
Lag
12
24
36
Chi-Square
9.6(DF=10)
17.0(DF=22)
* (DF= *)
48
* (DF= *)

Period
34
35
36
37
38
Forecast
3.1362
3.5810
3.8708
4.0594
4.1823
95 Percent Limits
Lower
Upper
0.3216
5.9507
-1.8180
8.9801
-2.3061
10.0476
-2.4192
10.5380
-2.4201
10.7848
zt = 1.54 + 0.65zt 1 + at 0.99at 1 , at : N ( 0, 2.06 )
Actual
( )
= 0.9857, s.e. ( ) = 0.0516, t = 19.11
= 1.5385, s.e. ( ) = 0.4894, t = 3.14
1 = 0.6513, s.e. 1 = 0.1434, t = 4.54
1
2 = 2.0613, with d . f . = 30
192
ACF of Residuals for CPIChnge

1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
1
Lag
PACF of Residuals for CPIChnge

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
1
Lag
193

(response is CPIChnge)
8
7
Frequency
6
5
4
3
2
1
0
-3
-2
-1
Residual
.
:

(response is CPIChnge)
4
3
Residual
2
1
0
-1
-2
-3
-2
-1
Normal Score
.
. 5
194
Time Series Plot for CPIChnge

(with forecasts and their 95% confidence limits)
CPIChnge
10
10
15
20
25
30
Time
:( AR(2
MTB > arima 2 0 0 c2
Type
AR
1
AR
2
Constant
Mean
Coef
1.1872
-0.4657
1.3270
4.765
StDev
0.1625
0.1624
0.2996
1.076
T
7.31
-2.87
4.43
Residuals:
MS = 2.9540 DF = 30
Lag
12
24
36
Chi-Square
19.8(DF=10)
25.4(DF=22)
* (DF= *)
zt = 1.33 + 1.187 zt 1 0.4657 zt 1 + at , at : N ( 0, 2.95)
48
* (DF= *)
( )
= 0.4657, s.e. ( ) = 0.1624, t = 2.87
= 1.327, s.e. ( ) = 0.2996, t = 4.43
1 = 1.1872, s.e. 1 = 0.1625, t = 7.31
2
2 = 2.954, with d . f . = 30
195
H 0 : 2 = 0
H 1 : 2 0
2
0.4657
=
= 2.8676
0.1624
s.e. 2
) (
P-value
= t0
;MTB > cdf -2.8676

SUBC> t 30.
Cumulative Distribution Function

Student's t distribution with 30 DF
)P( X <= x
0.0037
x
-2.8676
P-value 0.0037 0.05

2 = 0 . (AR(2
:

. AIC
:

-103.2
-391.4
-836.1
-1307.3
-1736.9
-2097.5
-2363.4
-2721.4
-3353.3
-4153.0
-5068.7
-5848.8
-6387.3
-6834.5
-7259.9
-7891.4
-8791.2
-9878.4
-11071.9
-12157.5
-13190.3
-14196.3
-15335.9
-16697.1
-18155.6
-19547.0
-20827.7
-21908.3
-76.7
-339.9
-766.7
-1242.9
-1679.0
-2055.4
-2328.5
-2651.9
-3253.4
-4028.2
-4937.2
-5760.6
-6317.8
-6773.3
-7193.3
-7784.7
-8649.8
-9713.3
-10903.2
-12011.7
-13039.8
-14051.7
-15158.3
-16492.7
-17951.5
-19356.3
-20655.7
-21762.1
-52.8
-291.3
-698.2
-1177.4
-1620.8
-2010.0
-2290.6
-2589.5
-3156.1
-3906.6
-4805.0
-5663.8
-6244.3
-6711.8
-7131.7
-7683.2
-8512.3
-9552.0
-10734.0
-11863.9
-12889.7
-13910.1
-14986.4
-16290.9
-17745.2
-19161.7
-20478.0
-21614.0
)z(t
-33.1
-246.0
-631.6
-1111.3
-1561.5
-1960.6
-2250.5
-2533.0
-3060.8
-3788.7
-4673.3
-5558.0
-6167.9
-6649.0
-7073.1
-7586.4
-8379.1
-9394.6
-10564.0
-11713.8
-12740.8
-13769.9
-14819.8
-16091.5
-17535.7
-18965.1
-20294.5
-21463.7
-22474.4
196
-19.2
-204.0
-566.7
-1044.2
-1500.0
-1906.9
-2210.8
-2482.0
-2968.6
-3675.0
-4542.3
-5444.4
-6090.4
-6584.6
-7015.6
-7495.3
-8249.2
-9239.6
-10392.1
-11560.2
-12593.4
-13629.3
-14657.5
-15895.8
-17324.4
-18766.2
-20107.3
-21310.9
-22335.1
-9.1
-165.9
-504.8
-975.1
-1436.5
-1851.1
-2173.3
-2437.6
-2880.4
-3564.3
-4412.0
-5323.6
-6011.8
-6518.9
-6957.1
-7411.2
-8124.0
-9086.3
-10219.0
-11402.0
-12447.5
-13486.3
-14499.7
-15705.0
-17113.5
-18564.2
-19919.5
-21154.4
-22195.6
-2.5
-132.4
-446.4
-905.3
-1371.8
-1794.5
-2136.4
-2398.7
-2797.6
-3456.9
-4281.7
-5197.8
-5931.6
-6453.5
-6896.0
-7332.9
-8004.6
-8936.4
-10047.2
-11238.9
-12302.5
-13339.6
-14345.9
-15518.4
-16904.2
-18360.1
-19733.5
-20993.7
-22053.2
:
Original Time Series
-10000
-20000
50
100
150
200

Autocorrelation
Index
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
10
Lag Corr
1
2
3
4
5
6
7
8
9
10
11
12
LBQ
0.9813.93 196.91
0.97 8.00 388.73
0.95 6.15 575.44
0.94 5.15 757.06
0.92 4.50 933.59
0.91 4.04
1105.06
0.89 3.68
1271.47
0.88 3.39
1432.85
0.86 3.15
1589.23
0.84 2.95
1740.66
0.83 2.78
1887.17
0.81 2.62
2028.81
Lag Corr
13
14
15
16
17
18
19
20
21
22
23
24
0.80
0.78
0.76
0.75
0.73
0.72
0.70
0.68
0.67
0.65
0.64
0.62
20
T
LBQ
2.49
2165.64
2.37
2297.73
2.25
2425.14
2.15
2547.94
2.06
2666.22
1.97
2780.05
1.89
2889.51
1.82
2994.69
1.75
3095.67
1.68
3192.54
1.62
3285.39
1.56
3374.32
30
Lag Corr
25
26
27
28
29
30
31
32
33
34
35
36
0.61
0.59
0.58
0.56
0.55
0.53
0.52
0.50
0.49
0.47
0.46
0.45
LBQ
1.50
3459.42
1.45
3540.78
1.40
3618.50
1.35
3692.67
1.30
3763.38
1.26
3830.74
1.21
3894.84
1.17
3955.76
1.13
4013.61
1.09
4068.47
1.05
4120.44
1.01
4169.59
40
Lag Corr
37
38
39
40
41
42
43
44
45
46
47
48
0.43
0.42
0.41
0.39
0.38
0.36
0.35
0.34
0.32
0.31
0.30
0.28
LBQ
50
Lag Corr
0.98
4216.02
0.94
4259.81
0.91
4301.04
0.87
4339.79
0.84
4376.14
0.81
4410.16
0.77
4441.93
0.74
4471.52
0.71
4498.99
0.68
4524.43
0.65
4547.90
0.62
4569.48
LBQ
49 0.27 0.59
4589.24
50 0.26 0.56
4607.24

z(t)
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag PAC
1
2
3
4
5
6
7
8
9
10
11
12
0.98
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
10
T
13.93
-0.17
-0.17
-0.16
-0.16
-0.16
-0.15
-0.15
-0.15
-0.14
-0.14
-0.13
20
Lag PAC
13
14
15
16
17
18
19
20
21
22
23
24
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
T
-0.13
-0.12
-0.12
-0.11
-0.11
-0.11
-0.11
-0.10
-0.10
-0.09
-0.09
-0.09
197
Lag PAC
25
26
27
28
29
30
31
32
33
34
35
36
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
30
T
-0.09
-0.09
-0.09
-0.08
-0.08
-0.08
-0.08
-0.08
-0.08
-0.08
-0.09
-0.09
40
Lag PAC
37
38
39
40
41
42
43
44
45
46
47
48
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
-0.01
50
Lag PAC
-0.09
-0.09
-0.10
-0.10
-0.11
-0.11
-0.12
-0.12
-0.12
-0.13
-0.13
-0.13
49 -0.01
50 -0.01
-0.14
-0.14
. zt
wt = zt zt 1
First Differences w(t)=z(t)-z(t-1)
w(t)
-100
-200
Index
50
100
150
200
:
Autocorrelation

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag Corr
1
2
3
4
5
6
7
8
9
10
11
12
LBQ
0.9913.93197.10
0.97 7.98389.03
0.95 6.12574.74
0.93 5.10753.55
0.91 4.43924.81
0.89 3.94
1087.93
0.86 3.56
1242.44
0.83 3.25
1388.10
0.81 2.98
1524.83
0.78 2.76
1652.58
0.75 2.56
1771.42
0.72 2.38
1881.49
15
Lag Corr
13
14
15
16
17
18
19
20
21
22
23
24
0.69
0.66
0.63
0.60
0.57
0.54
0.52
0.49
0.47
0.45
0.43
0.41
LBQ
2.21
1983.03
2.07
2076.38
1.93
2161.98
1.81
2240.33
1.70
2311.99
1.60
2377.42
1.50
2437.13
1.41
2491.57
1.33
2541.23
1.26
2586.61
1.19
2628.22
1.13
2666.48
25
Lag Corr
25
26
27
28
29
30
31
32
33
34
35
36
0.39
0.38
0.37
0.36
0.35
0.34
0.34
0.34
0.34
0.34
0.34
0.34
LBQ
1.08
2701.91
1.03
2734.99
1.00
2766.16
0.96
2795.84
0.94
2824.46
0.92
2852.35
0.91
2879.79
0.90
2907.03
0.89
2934.32
0.89
2961.85
0.89
2989.82
0.89
3018.34
35
Lag Corr
37
38
39
40
41
42
43
44
45
46
47
48
0.34
0.35
0.35
0.35
0.35
0.35
0.35
0.34
0.34
0.33
0.33
0.32
45
LBQ
0.90
3047.50
0.90
3077.32
0.90
3107.68
0.90
3138.42
0.90
3169.33
0.89
3200.21
0.88
3230.81
0.87
3260.99
0.85
3290.64
0.84
3319.59
0.82
3347.55
0.79
3374.17
Lag Corr
LBQ
49 0.31 0.76
3399.16

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
Lag PAC
1
2
3
4
5
6
7
8
9
10
11
12
0.99
-0.14
-0.11
-0.07
-0.06
-0.07
-0.06
-0.03
-0.03
-0.03
-0.02
-0.02
15
Lag PAC
13.93
-1.97
-1.51
-0.96
-0.90
-0.97
-0.78
-0.48
-0.37
-0.37
-0.34
-0.22
13 -0.01
14 -0.00
15 0.01
16 0.02
17 0.02
18 0.01
19 0.01
20 0.01
21 0.01
22 0.03
23 0.02
24 0.02
-0.16
-0.02
0.10
0.23
0.25
0.10
0.12
0.11
0.16
0.35
0.31
0.27
25
Lag
PAC
25 0.04
26 0.04
27 0.03
28 0.04
29 0.04
30 0.02
31 0.02
32 0.01
33 0.01
34 0.01
35 0.00
36 -0.00
0.52
0.62
0.47
0.55
0.56
0.34
0.22
0.16
0.18
0.15
0.04
-0.07
35
Lag PAC
37
38
39
40
41
42
43
44
45
46
47
48
-0.01
-0.02
-0.04
-0.04
-0.04
-0.04
-0.03
-0.00
0.01
-0.01
-0.04
-0.05
45
T
Lag PAC
-0.07
-0.29
-0.63
-0.60
-0.54
-0.55
-0.42
-0.07
0.08
-0.18
-0.56
-0.76
49 -0.03
-0.48
. wt
) ( yt = wt wt 1
198
First Differences y(t)=w(t)-w(t-1)

10
y
(t)
-5
Index
50
100
150
200
:
Autocorrelation
Autocorrelation Function for y(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag Corr
1
2
3
4
5
6
7
8
9
10
11
12
LBQ
0.9313.08173.62
0.80 6.83303.36
0.70 4.89401.81
0.62 3.89479.62
0.56 3.26542.93
0.50 2.77593.77
0.43 2.32632.35
0.36 1.90659.84
0.29 1.50677.66
0.22 1.11687.71
0.15 0.75692.41
0.08 0.41693.82
15
Lag Corr
LBQ
25
Lag Corr
13 0.01 0.06693.85
14 -0.05 -0.23694.31
15 -0.09 -0.46696.12
16 -0.14 -0.70700.34
17 -0.19 -0.97708.51
18 -0.25 -1.24722.16
19 -0.31 -1.53743.30
20 -0.37 -1.83774.40
21 -0.44 -2.13818.29
22 -0.51 -2.40876.73
23 -0.56 -2.57948.00
24 -0.60 -2.67
1030.79
LBQ
25 -0.64 -2.73
1124.21
26 -0.66 -2.72
1224.05
27 -0.66 -2.62
1324.28
28 -0.63 -2.44
1418.06
29 -0.59 -2.22
1500.63
30 -0.55 -2.01
1572.45
31 -0.50 -1.80
1632.42
32 -0.43 -1.53
1677.44
33 -0.36 -1.25
1708.46
34 -0.29 -1.00
1728.63
35 -0.22 -0.76
1740.51
36 -0.16 -0.53
1746.42
35
Lag Corr
45
LBQ
37 -0.09 -0.30
1748.28
38 -0.01 -0.03
1748.29
39 0.08 0.26
1749.70
40 0.15 0.50
1755.14
41 0.20 0.68
1765.20
42 0.23 0.78
1778.55
43 0.24 0.82
1793.44
44 0.25 0.86
1810.08
45 0.28 0.96
1831.09
46 0.34 1.13
1860.92
47 0.40 1.34
1903.36
48 0.45 1.49
1956.81
Lag Corr
LBQ
49 0.48 1.56
2017.77
Partial Autocorrelation Function for y(t)

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
Lag PAC
1
2
3
4
5
6
7
8
9
10
11
12
0.93
-0.46
0.33
-0.14
0.12
-0.13
0.00
-0.06
-0.08
-0.04
-0.04
-0.09
T
13.08
-6.44
4.71
-1.97
1.72
-1.78
0.01
-0.81
-1.11
-0.55
-0.58
-1.21
15
Lag PAC
13
14
15
16
17
18
19
20
21
22
23
24
-0.05
0.05
-0.07
-0.09
-0.04
-0.08
-0.12
-0.11
-0.15
-0.07
-0.06
-0.15
T
-0.75
0.67
-0.95
-1.30
-0.55
-1.11
-1.71
-1.50
-2.12
-0.93
-0.87
-2.12
25
Lag PAC
25
26
27
28
29
30
31
32
33
34
35
36
-0.03
-0.03
0.00
0.08
-0.01
-0.04
0.15
0.06
-0.00
0.04
0.03
0.04
T
-0.46
-0.44
0.04
1.13
-0.14
-0.56
2.17
0.88
-0.05
0.49
0.46
0.50
35
Lag PAC
37
38
39
40
41
42
43
44
45
46
47
48
0.08
0.11
0.04
0.04
-0.05
-0.05
-0.14
-0.01
0.07
0.06
-0.01
-0.09
45
T
Lag PAC
1.16
1.56
0.51
0.52
-0.74
-0.66
-1.95
-0.10
0.92
0.79
-0.17
-1.24
49 0.07
1.00

. d=2

zt ( ARIMA(1,2,1
MTB > ARIMA 1 2 1 'z(t)' 'RESI2' 'FITS2';
SUBC>
NoConstant;
SUBC>
SUBC> GACF;
SUBC> GPACF;
SUBC> GHistogram;
199
SUBC>
GNormalplot.
ARIMA Model
Iteration
SSE
Parameters
0
2462.77
0.100
0.100
1
1345.58
0.250
-0.050
2
1170.63
0.203
-0.200
3
984.83
0.182
-0.350
4
782.47
0.200
-0.500
5
560.15
0.278
-0.650
6
363.93
0.428
-0.765
7
259.20
0.578
-0.814
8
202.76
0.728
-0.842
9
185.51
0.861
-0.859
10
185.36
0.873
-0.860
11
185.36
0.875
-0.860
12
185.36
0.875
-0.860
Type
Coef
StDev
AR
1
0.8749
0.0353
MA
1
-0.8599
0.0357
0.0010
T
24.75
-24.12
Differencing: 2 regular differences

Residuals:
MS =
0.937 DF = 196
Lag
12
24
36
Chi-Square
3.9(DF=10)
13.0(DF=22)
33.1(DF=34)
48
46.0(DF=46)

Period
201
202
203
204
205
206
207
208
209
210
Forecast
-22615.2
-22757.2
-22900.4
-23044.5
-23189.5
-23335.2
-23481.5
-23628.4
-23775.8
-23923.7
95 Percent Limits
Lower
Upper
-22617.1
-22613.3
-22764.6
-22749.9
-22917.2
-22883.5
-23075.3
-23013.7
-23238.8
-23140.1
-23407.8
-23262.6
-23582.1
-23380.9
-23761.8
-23495.0
-23946.7
-23605.0
-24136.6
-23710.7
zt = 0.875t 1 z + at 0.859at 1 , at : N ( 0,0.937 )
Actual
200
( )
s.e. ( ) = 0.0357,
1 = 0.8749, s.e. 1 = 0.0353, t = 24.75

1 = 0.8599,
t = 24.12
2 = 0.937, with d . f . = 196
.
:
1.0
0.8
Autocorrelation
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
Lag

1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
5
10
15
20
25
30
35
40
45
Lag
201

))(response is z(t
20
Frequency
10
0
1
-1
-3
-2
Residual
.
:
))(response is z(t
-1
-2
-3
3
-1
-2
-3
Normal Score
.
10 %95
.
202
Residual
Forecast of 20 Future values with 95% limits

-22500
z(t)
-23000
-23500
-24000
0
10
Time
203
:
1-
Abraham, B. and Ledoter, J. (1983). Statistical

Methods for Forecasting, John Wiley, New York.
2Anderson, T. W. (1971). The Statistical Analysis of
Time Series, John Wiley, New York.
345678-
Box, G. E. P. and Jenkins, G. M. (1976). Time Series Analysis

Forecasting and Control, 2nd ed., Holden-Day, San Francisco.
Brillinger, D. R. (1975). Time Series: Data Analysis and
Theory, Holt, Rinehart and Winston, New York.
Fuller, W. A. (1976). Introduction to Statistical Time Series,
John Wiley, New York.
Granger, C. W. J. and Newbold, P. (1977). Forecasting
Economic Time Series, Academic Press, New York.
Hannan, E. J. (1970). Multiple Time Series, John Wiley, New
York.
Harvey, A. C. (1980). Time Series Models, Halsted Press, New
York.
9-
Montgomery, D. C., Johnson, L. A. and Gardiner, J. S.

(1990). Forecasting and Time Series Analysis, 2nd ed.,
McGraw-Hill International Edition.
10Makridakis, S., Wheelwright, S. C. and McGee, V.
E. (1983). Forecasting Methods and Applications, 2nd ed.,
John Wiley, New York.
11Shumway, R. H. (1988). Applied Statistical Time Series
Analysis, Prentice-Hall, New York.
12Wei, W. W. S. (1990). Time Series Analysis Univariate and
Multivariate Methods, Addison Wesley.
13Minitab Reference Manual, Release 11 for Windows.
(1998).

:
abarry@ksu.edu.sa abarry@abarry.net
.
204

Statistical Forecasting Methods

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Statistical Forecasting Methods

Uploaded by

Copyright:

Available Formats

) (

5.3.2.2 AIC ......................................... 80

:6 } { z1 , z2 ,L, zn1 , zn Stationary

:10 (Partial Autocorrelation Function (PACF

: 12 Sample Autocorrelation Function SACF

= 0.05 H 0 n rk > 1.96

:13 Sample Partial Autocorrelation

rkj = rk 1, j rkk rk 1,k 1 ,

= 0.05 H 0 n rkk > 1.96

) ( 158 222 + 222 248 + L + 178 169

r8 = 0.230, r7 = 0.104, r6 = 0.242, r5 = 0.387, r4 = 0.183

V ( r4 ) 0.138 V ( r5 ) 0.1454 V ( r6 ) 0.1787

rkj = rk 1, j rkk rk 1,k 1 ,

r2 r11r1 ( 0.212 ) ( 0.265) ( 0.265) 0.282225

( 0.076 ) ( ( 0.345) ( 0.212 ) + ( 0.303) ( 0.265) )

Autocorrelation Function for Demand

Partial Autocorrelation Function for Demand

Autocorrelation function of AR(1) Model

Partial Autocorrelation function of AR(1) Model

11 = 0.944186 22 = -0.384471 33 = 0.183710

wt = { b0 2b0 + b0 } + { b1t 2b1 ( t 1) + b1 ( t 2 ) } +

Differenced and Transformed Series

: - ) (1,1 : (ARIMA(1,1,0) =ARI(1,1

: Random Walk with Trend Model

zt = at + 1at 1 + 2at 2 + 3at 3 + L

zt = at + 1at 1 + 2at 2 + 3at 3 + L

E zn + l zn ( l) = E an + l + 1an + l1 + L + l1an +1 + ( l 0 ) an + ( l+1 1 ) an 1 + L

en ( l) = zn + l zn ( l) = an +l + 1an +l1 + 2an +l2 + L + l1an +1 , l 1

zn ( l) = lan + l+1an 1 + l+ 2an 2 + L, l 1

zn ( l) = lan + l+1an 1 + l+ 2an 2 + L , l 1

= +1E ( zn + l1 zn , zn1 ,L) + E an + l zn , zn 1 ,L , l 1

l = 1: zn ( 1) = +1E ( zn zn , zn 1 ,L) + E an +1 zn , zn 1,L

l = 2 : zn ( 2 ) = +1E ( zn +1 zn , zn 1 ,L) + E an + 2 zn , zn 1,L

zt 0.97 = 0.85 ( zt 1 0.97 ) + at , at : WN ( 0,0.024 )

= 0.97+0.85 ( 0.49 0.97 ) = 0.56

z156 ( 2 ) = 0.97+0.85 ( z156 ( 1) 0.97 )

= 0.97+0.85 ( 0.56 0.97 ) = 0.62

z156 ( 3) = 0.97+0.85 ( z156 ( 2 ) 0.97 )

= 0.97+0.85 ( 0.62 0.97 ) = 0.68

2 1 < 1, 2 + 1 < 1, 2 < 1

= +1E ( zn + l1 zn , zn1 ,L) + 2 E ( zn + l 2 zn , zn 1 ,L) + E an + l zn , zn 1,L , l 1

zn ( l) = +1E ( zn + l1 zn , zn 1 ,L) + 2 E ( zn + l2 zn , zn 1 ,L) + E an + l zn , zn 1 ,L , l 1

l = 1: zn ( 1) = +1E ( zn zn , zn 1 ,L) + 2 E ( zn 1 zn , zn 1,L) + E an +1 zn , zn 1 ,L

l = 3 : zn ( 3) = +1E ( zn + 2 zn , zn 1 ,L) + 2 E ( zn +1 zn , zn 1,L) + E an +3 zn , zn 1 ,L

l = 4 : zn ( 4 ) = +1E ( zn +3 zn , zn 1 ,L) + 2 E ( zn + 2 zn , zn 1,L) + E an + 4 zn , zn 1 ,L

=E ( zn + l1 zn , zn 1 ,L) + E ( an +l zn , zn 1 ,L) 1E ( an+ l1 zn , zn 1 ,L) , l 1

zn ( l) =E ( zn + l1 zn , zn 1 ,L) + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) , l 1

l = 1: zn ( 1) =E ( zn zn , zn 1 ,L) + E ( an+1 zn , zn 1 ,L) 1E ( an zn , zn1 ,L)

l = 2 : zn ( 2 ) =E ( zn +1 zn , zn 1 ,L) + E ( an + 2 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L)

l = 3 : zn ( 3) =E ( zn + 2 zn , zn 1 ,L) + E ( an +3 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L)

= + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) , l 1

= + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) 2 E ( an +l2 zn , zn 1,L) , l 1

zn ( l) = + E ( an + l zn , zn 1 ,L) 1E ( an + l1 zn , zn 1 ,L) 2 E ( an + l2 zn , zn 1 ,L) , l 1

l = 1: zn ( 1) = + E ( an +1 zn , zn 1 ,L) 1E ( an zn , zn 1 ,L) 2 E ( an 1 zn , zn 1 ,L)

l = 2 : zn ( 2 ) = + E ( an + 2 zn , zn 1 ,L) 1E ( an +1 zn , zn 1 ,L) 2 E ( an zn , zn 1 ,L)

l = 3 : zn ( 3) = + E ( an +3 zn , zn 1 ,L) 1E ( an + 2 zn , zn 1 ,L) 2 E ( an +1 zn , zn 1 ,L)

= + 1E ( zn + l1 ) zn , zn 1 ,L + E ( an + l zn , zn 1 ,L) 1E ( an +l1 zn , zn 1 ,L) , l 1

l = 1: zn ( 1) = + 1E ( zn ) zn , zn 1 ,L + E ( an +1 zn , zn1 ,L) 1E ( an zn , zn 1,L)

l = 2 : zn ( 2 ) = + 1E ( zn +1 ) zn , zn 1 ,L + E ( an + 2 zn , zn1 ,L) 1E ( an+1 zn , zn 1 ,L)

= 0.05 . u0.025 = 1.96

z156 ( 1) = 0.56, z156 ( 2 ) = 0.62, z156 ( 3) = 0.68

V e156 ( 1) = 0.024, V e156 ( 2 ) = 0.041, V e156 ( 2 ) = 0.054

%95 z157 , z158 , z159 21

= 0.56 1.96 0.024 = 0.56 0.304