.


:




:

































 

 

 

 





p q .

1. ( 7.2) Transformations of Variables ( ) Autocorrs ().

2. Autocorrelations (), , p-level for highlighting. 7.6.

Partial autocorrelations, 7.7.

 

 

7.5 , ( )

 

7.6

 

7.7

 

7.1 ARMA . ( ) ARMA (1,0) ARMA (2,0). ( 7.7, ) . , x=f(x) Autocorr. (x=x-(a+b*x(lag))), ( ) x-0,000-1,00*x(t-1)

 

 

7.8 ( )

:

 

 

7.9

 

 

7.10

 

ARMA(1,0), - , - () 1. .

ARMA

 

:

1) ARMA Multiple regression

2) ARMA ARIMA & Autocorrelation function

7.7.1. ARMA Multiple regression

 

ARMA ( ). :

1. , Transformations of Variables Save variables , :

Y 0;

Y_1 ;

Y_2 .

2. : Y_1 D, Y_2 D.

Dt-1, Insert Add Variables. Add Variables Name NewVar Dt-1, Long name = v3.

3. Date Shift (Lag) Dt-1 (Forward).

4. Statistics Multiple Regression. Select dependent and independent variable lists ( Variables) D - Dt-1.

5. Advanced options (stepwise or ridge regression) Model Definition Intercept Set to zero (.. ). :

 

7.2

 

  Beta Std.Err. of Betta B Std.Err. of B t(24) p-level
D''t-1 -0,878 0,098 -0,911 0,101 -8,992 0,000

 

ARMA(1,0), a . , F - (F (1,24)=80,86 F (1,24)= 4,26).

6. (Multiple Regression Results) Residuals / assumptions / prediction, .

Residual Analysis Save Save residuals & predicted, Select variables to save with predicted/residual sc D . :

 

 

7.11 ARMA

 

7. . Graphs 2D Graphs Line Plots (Variables). Variables () D Predicted. Graph type ( 2D Line Plots Variables) Multiple ( , ).

 

 

7.12 - 0

ARMA (1,0) .

8. Residuals/assumptions/prediction Predict dependent variable .

9. Specify values for indep. vars Dt-1 -79, :

 

7.3 0 1 2006.

 

  B-Weight Value B-Weight * Value
D''t-1 -0,911 -79,000 71,969
Predicted     71,969
-95,0%CL     55,451
+95,0%CL     88,488

 

2-4 2006. :

 

7.4 - 0 2006.

 

/ ARMA (1,0)
I/2006 -79,000 71,969 55,451 88,488
II/2006 71,969 -65,564 -80,612 -50,516
III/2006 -65,564 59,729 46,020 73,438
IV/2006 59,729 -54,413 -66,902 -41,924

7.7.2. ARMA ARIMA & Autocorrelation function

 

ARMA STATISTICA :

1. Statistics Advancer Linear/Nonlinear Models Time Series Analysis ARIMA & Autocorrelation function ( ARIMA & ).

2. Single Series ARIMA ( ) Quick (). , , x-0,000-1,00*x(t-1); x-0,000-1,00*x(t-1) ( ).

 

 

7.13 ARMA ( )

 

3. Single Series ARIMA Results ( ).

 

 

7.14

 

( 7.14) , : ARMA , , .

4. Advanced Summary: Parameter estimates 7.5.

 

7.5 ARMA(1,0)

 

  Param. Asympt. Std. Err. Asympt. t(25) p Lower 95% Conf Upper 95% Conf
p(1) -0,911 0,102 -8,895 0,000 -1,122 -0,700

 

:

.

t -.

p- 95% () .

t - .

5. Advanced Forecasting (). 4 , 4 Number of cases ( ). Forecast cases ( ), 7.6 7.15.

 

7.6 - 0 2006.

 

  Forecast Lower 95% Upper 95% Std.Err.
  -162,614 -291,707 -33,522 62,680
  148,142 -26,487 322,772 84,791
  -134,958 -339,833 69,917 99,476
  122,947 -103,989 349,884 110,188

 

 

7.15 ARMA (1,0)

 

 

1) () , () , :

) ARMA(0,2)

) ARMA(0,1)

) ARMA(1,0)

2) , , :

) ARMA(0,2)

) ARMA(0,1)

) ARMA(1,0)

3) , , :

) ARMA(2,0)

) ARMA(0,1)

) ARMA(1,0)

4) () 1 2, , :

) ARMA(0,2)

) ARMA(0,1)

) ARMA(1,0)

5) yt = ayt-1 + et :

) AR(1)

) AR(2)

) MA(1)

) MA(2)

6) yt = a1yt-1 + a2yt-2 + et :

) AR(1)

) AR(2)

) MA(1)

) MA(2)

7) yt = et +qet-1 :

) AR(1)

) AR(2)

) MA(1)

) MA(2)

8) yt = et +q1et-1 + q2et-2 :

) AR(1)

) AR(2)

) MA(1)

) MA(2)

9) yt = a1yt- + et +qet-1 :

) ARIMA (1, 1, 1)

) ARMA (1, 1)

) ARMA (2, 2)

10) :

) yt = ayt-1 + et

) yt = a1yt-1 + a2yt-2 + et

) yt = a1yt- + et +qet-1

 





:


: 2016-11-02; !; : 2398 |


:

:

.
==> ...

1382 - | 1363 -


© 2015-2024 lektsii.org - -

: 0.166 .