ARIMA

:
ARIMA

Autoregressive
Integrated Moving Average .

Box-Jenkins AR

. MA

) 70 (

STATGRAPHICS

) ARIMA(1,0,1
) ARIMA (1,0,0 .

) (
)(
2001 . 2004 -
/ .

-1

.

.

.
.
.
: :

.
) Autoregressive integrated moving average (ARIMA
. Box
Jenkins 1970 Box-Jenkins

)(1

AR . MA
: :

ARIMA

)(2

Shhui Deng and Bin Liu

)(3

James Hansen , James Mcdonald and Ray Nelson

ARIMA
)(4

Neural Network

Jeffery Wanger

.
:

ARIMA

STATGRAPHICS

. :
.ARIMA
1

.
AR .MR

.ARIMA
.

-2 ARIMA
ARIMA George Box
Gwilyn Jenkins 1970
:

- AR

)(1

)(5

YT = B0 + B1YT 1 + B2YT 2 + ...... + BPYT P + eT

:
: YT Y .
: YT-1,YT-2,YT-P Y T
: B0,B1,B2,Bp

YT
YT-P YT-2 YT-1

- MA

)(2

)(6

YT = W o + eT W1eT 1 W 2 eT 2 ....... W q eT q

: YT Y .
: eT-1,eT-2,eT-q YT
: W0, W1, W2, Wq
: eT

 YT .eT-1,eT-2,eT-q
ARMA
)autoregressive (ARMA

moving average :

YT = B0 + B1YT 1 + B2YT 2 + ...... + BPYT P + eT

)(3

+ Wo + eT W1eT 1 W2 eT 2 ....... Wq eT q

ARMA ) (p,q p

q .

-3 :
Box-Jenkins
stationary
) .(7

Price and Box :
)(4

: Q Price and Box
: T

: X2
: K
Q K

) (K )
( :

3

 .
.
differencing

YT YT-1 YT-1 YT-2

d 1
.

-4 AR MA
AR MA
.

. AR
MA :
) (1 ARMA

ACF

PACF

)AR(1

)AR(2

)AR(p

)MR(1

)MR(2

)MR(q

)ARMA(1,1

)ARMA(p,q

):(8

- mean absolute error :

MAE = |eT| / n
eT = YT-FT

)(5
)(6
:

: eT
: YT
: FT YT

- mean squared error

:

eT2 / n

)(7

= MSE

- mean absolute percentage error
:
MAPE = (|eT|/YT )/ n

)(8

- mean percentage error

:

(eT2/YT )/ n

)(9

= MPE

 -6 ARIMA
70
) (2 70

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

47
64
23
71
38
64
55
41
59
48
71
35
57
40
58
44
80
55
37
74
51
57
50
60
45

57
50
45
25
59
50
71
56
74
50
58
45
54
36
54
48
55
45
57
50
62
44
64
43
52

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
61
62
63
64
65
66
67
68
69
70

38
59
55
41
53
49
34
35
54
45
68
38
50
60
39
59
40
47
54
23

ARIMA :
:
)
STATGRAPHICS :

 ) (1

Estimated Autocorrelations for prod

1

0.2
-0.2
-0.6

Autocorrelations

0.6

-1
24

20

16

12

lag
) (2

Estimated Partial Autocorrelations for prod

0.6
0.2
-0.2
-0.6
-1
24

20

16

12

Partial Autocorrelations

lag

) ARIMA(1,0,1 )ARIMA(1,0,0

STATGRAPHCS :
:
ARIMA(1,0,1)
Estimation
Statistic
Period
-------------------------MSE
118.193
MAE
7.98234
MAPE
18.2879
MPE
-5.636
ARIMA Model Summary
Parameter
Estimate
Stnd. Error
t
P-value
---------------------------------------------------------------------------AR(1)
-0.7159
0.186062
-3.84765
0.000269
MA(1)
-0.357292
0.249327
-1.43303
0.156501
Mean
51.1129
1.03326
49.4677
0.000000
Constant
87.7046
----------------------------------------------------------------------------

ARIMA(1,0,0)
Estimation
Statistic
Period
---------------------------MSE
120.751
MAE
8.23681
MAPE
18.7248
MPE
-5.73502
ARIMA Model Summary
Parameter
Estimate
Stnd. Error
t
P-value
---------------------------------------------------------------------------AR(1)
-0.414269
0.115931
-3.57342
0.000653
Mean
51.1122
0.939391
54.4099
0.000000
Constant
72.2864
----------------------------------------------------------------------------

:
ARIMA(1,01)
( p value=0.000256) AR(1)
.(P value = 0.156) MR(1) (0.05)

ARIMA(1,0,0)

.(P value=0.00065) AR(1)

ARIMA(1,1,0)

( 4)( 3)
ARIMA(1,0,0)

)ARIMA (1,0,1
90

Data
Forecast

80
70
60
50
40
30
20
10
0

9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69

)(

) (3 )ARIMA(1,0,1

)ARIMA(1,0,0
Data
90

Forecast

80
70
60
50
40
30
20
10
0
9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69

)(

) (4 )ARIMA(1,0,0

 : ) ARIMA(1,0,1
) ARIMA(1,0,0 .
:
) (2 ARIMA

MSE
MAE
MAPE
MPE

)ARIMA(1,0,1
118.193
7.98234
18.2879
-5.636

)ARIMA(1,0,0
120.751
8.23681
18.7248
-5.73502

-7
ARIMA

. AR

MA

. Autoregressive Integrated Moving Average

ARIMA
:
.1
.

.2 .
.3 .
.4 .
ARIMA
) ARIMA (1,0,1 ) ARIMA(1,0,0
.

10

(1 ) Box G, Jenkins G, (1970), Time Series Analysis, Forecasting and

Control, San Francisco: Holden-Day
( 2)Suhui Deng and Bin Liu " Modeling and Forecasting demand for
money in China,(1999) :cointegration and nonlinear analysis"
Journal of Annals Operations Research Vol. 87 pp:177-189
( 3) James Hansen, James Mcdonald and Ray Nelson (1999)" Time series
prediction with genetic-algorithm designed neural network : an
empirical comparison with modern statistical models" Journal of
Computational Intelligence, Vol. 15 No. 3 , pp-171-184
( 4) Jeffery Wamger (2000)" Estimation the optimal scale of public
investment : the case low-level Radioactive waste disposal facilities "
Journal of Regulatory Economics , Vol. 17 No. 2, pp 173-188
( 5) John Hanke and Arthur Reitsch (1991)" Understanding Business
Statistics " Richard D. Irwn Inc, Boston, , P 718.
" " .( 6)
213 2002
(7 ) Regina Kaiser and Agustin Maravall " Notes on time series analysis
ARIMA models and Signal Extraction " Banco , Spanish , without
date. p 6
: ( 8)
Gerald Keller and Brian Worrack( 1997)" Statistics for Management
and Economics " Cole publishing Company , New York , , p 923
David Anderson, Dennis Sweeney and Thomas William "
Quantitative Methods for Business (2001)" South Western college
Publishing , Ohio, , p 173
Donald Harnett and James Horriel (1998)" Data, Statistics and
decision models with Excel ' John and sons, New York, , p 368

11