I have some fitted ARIMA model:

> fit
Series: mydata[1] 
ARIMA(1,0,1) with non-zero mean 
Box Cox transformation: lambda= 0.1088793 

         ar1      ma1     mean
      0.4536  -0.1702  -0.1868
s.e.  0.0094   0.0104   0.0033

sigma^2 estimated as 0.461:  log likelihood=-98549.78
AIC=197107.5   AICc=197107.5   BIC=197145.4

Now I want to check the prediction of this model. To do so, I want just to elongate some primer sequence


Thus, I use forecast() function:

fcast <- forecast(object = primer, model = fit)

My problems are:

1) I need only one value per step, but not mean

2) I am not sure, that this syntactic construction does what I need: use prior information only from primer and produces lambda adjusted results


I found, that I should use simulate function, and it will generate the sample paths conditional on the previous data. Thus I still have a question how to replace the data used for fitting by primer data?


You can apply a previously fitted ARIMA model to new data by using the Arima() function, feeding the new data into the y parameter and the fitted model into model. (Note the capitalization; this is a different function than arima()!)


Thus, you can simulate like this:


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.