Simulation of SARIMA

For technical support, tips and tricks, suggestions, or any other information regarding the EViews model object.

Moderators: EViews Gareth, EViews Moderator

mariusfu
Posts: 3
Joined: Wed Oct 27, 2010 4:21 am

Simulation of SARIMA

Postby mariusfu » Wed Oct 27, 2010 5:14 am

I have found a sarima model using R, and want to make a stochastic simulation of that model in Eviews 7.
The objective is to find a distribution of the values during the forecasting period.

I have:
- imported time series for the variable (inflow) and the residuals obtained in R
- created a model, based on the sarima from R:
d(inflow , 1 , 52) = - 0.3541*d(inflow(-1) , 1 , 52) + 1.1491 * d(inflow(-2) , 1 , 52) + 0.2523 * d(inflow(-3) , 1 , 52) - 0.2857 * d(inflow(-4) , 1 , 52) + 0.0078 * d(inflow(-5) , 1 , 52) - 0.0793 * d(inflow(-6) , 1 , 52) + 0.1148 * d(inflow(-7) , 1 , 52) + 0.0927 * residual_series(-1) - 1.7196 * residual_series(-2) - 0.1925 * residual_series(-3) + 0.8200 * residual_series(-4) - 0.8311 * residual_series(-52)

When solving the model, I have chosen simulation type = stochastic, solution sample=715 818, method=boothstrap, Boothstrap residual draw sample= 1 714

The result is unsuccessful and I get the message:
Unable to compute due to missing data in "D(INFLOW , 1 , 52) = - 0.3541 * D(INFLOW(-1) , 1 , 52) + 1. ...

It looks like the time series for the variable (inflow) is not used in the simulation.

Do anyone have any advice?
How do you recommend to simulate an already specified sarima model?

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 12266
Joined: Tue Sep 16, 2008 5:38 pm

Re: Simulation of SARIMA

Postby EViews Gareth » Wed Oct 27, 2010 8:22 am

Could you post your workfile?
Follow us on Twitter @IHSEViews

mariusfu
Posts: 3
Joined: Wed Oct 27, 2010 4:21 am

Re: Simulation of SARIMA

Postby mariusfu » Wed Oct 27, 2010 9:52 am

Here is my workfile.
Attachments
simulations1.wf1
workfile
(28.83 KiB) Downloaded 364 times

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 12266
Joined: Tue Sep 16, 2008 5:38 pm

Re: Simulation of SARIMA

Postby EViews Gareth » Wed Oct 27, 2010 10:21 am

Your specification has 0.0927*residual_series(-1). Residual_series only runs up until observation 702. Thus any value after observation 703 cannot be calculated.
Follow us on Twitter @IHSEViews

mariusfu
Posts: 3
Joined: Wed Oct 27, 2010 4:21 am

Re: Simulation of SARIMA

Postby mariusfu » Thu Oct 28, 2010 4:46 am

Instead I tried to estimate the model given coefficients from R as starting values.
I used the equation specification:
d(inflow,1,52) ar(1) ar(2) ar(3) ar(4) ar(5) ar(6) ar(7) ma(1) ma(2) ma(3) ma(4) sma(1)

I have chosen user supplied starting coefficient values in the option window for equation estimation.
How do I insert these values? Unsuccessfully I defined:
param ar(1) -0.3541 ar(2) 1.1491 ar(3) 0.2523 ar(4) -0.2857 ar(5) 0.0078 ar(6) -0.0793 ar(7) 0.1148 ma(1) 0.0927 ma(2) -1.7196 ma(3) -0.1925 ma(4) 0.8200 sma(1) -0.8311

Also, I want to use maximum likelihood estimation, but cannot find any place to specify that.
Do anyone have any advice for estimating the SARIMA(7,1,4,0,1,1) model?

Attached is my workfile:
Attachments
simulations2.wf1
workfile
(24.32 KiB) Downloaded 333 times

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 12266
Joined: Tue Sep 16, 2008 5:38 pm

Simulation of SARIMA

Postby EViews Gareth » Thu Oct 28, 2010 6:52 am

You can set the starting values by filling in the C vector in the workfile with those values.

EViews doesn't perform MLE estimation of ARIMA models.
Follow us on Twitter @IHSEViews


Return to “Models”

Who is online

Users browsing this forum: No registered users and 1 guest