how can i generate scenarios based on an adjusted ARIMA model?

5 ビュー (過去 30 日間)
Mazaher Haji Bashi
Mazaher Haji Bashi 2015 年 12 月 23 日
回答済み: Rakesh Chavan 2016 年 1 月 4 日
Dear all I have fitted a seasonal ARIMA model to my data and its fully working when i use forecast function. It means, the predicted 24 hours data completely conform the the original data behavior. The code is as follows:
mdl=regARIMA('ARLags',1,'SARLags',24,'MALags',1,'SMALags',[24 2*24],'D',1,'Seasonality',24,'Intercept',0);
[Estmdl,EstVar,logL,info]=estimate(mdl,L_Price)
and the forecasting code:
Day=3; T=Day*24; PER=24; v=forecast(Estmdl,PER,'Y0',L_Price(1:T,1)); figure plot(L_Price,'color',[0.70 0.70 0.70]) hold on plot(T+1:T+PER,v,'r') xlim([0,T+PER])
Actually L_Price is the log () of my original data to smooth the variations.
The problem is I want to generate scenarios for the data, facilitating to solve a stochastic optimization problem. I need at least 100 scenarios to be generated. When I use MATLAB simulate function, the things will go wrong (at least I think so) because the generated scenarios have not to do with that of the original data behavior. Does anyone has a solution? How can I generate the scenarios based on the seasonal ARIMA? Please advise.

回答 (1 件)

Rakesh Chavan
Rakesh Chavan 2016 年 1 月 4 日
Hi,
The 'simulate' function uses the estimated ARIMA model and the number of observations for each path. As your model is derived from the original observations, the behavior is contained in the model.
Is there any other error which you are referring when you use simulate?
-thanks

カテゴリ

Help Center および File ExchangeConditional Mean Models についてさらに検索

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by