In case further clarification is needed, this from Rob Hyndman, author
of the Forecast package, may be helpful:
"fitted produces one-step in-sample (i.e., training data) "forecasts".
That is, it gives a forecast of observation t using observations up to
time t-1 for each t in the data. ... So fitted(fit) gives one-step
forecasts of observations 1, 2, ... It is possible to produce a
"forecast" for observation 1 as a forecast is simply the expected value
of that observation given the model and any preceding history."
From Hyndman's answer in this thread
<a
href="https://stats.stackexchange.com/questions/217955/difference-between-first-one-step-ahead-forecast-and-first-forecast-from-fitted?rq=1">https://stats.stackexchange.com/questions/217955/difference-between-first-one-step-ahead-forecast-and-first-forecast-from-fitted?rq=1</a>
See also <a
href="https://robjhyndman.com/hyndsight/out-of-sample-one-step-forecasts/">https://robjhyndman.com/hyndsight/out-of-sample-one-step-forecasts/</a><https://robjhyndman.com/hyndsight/out-of-sample-one-step-forecasts/>
[A quick search on the stackexchange forum will turn up several similar
questions & answers]
HTH,
/
Best regards,
/
/
Partho Sarkar
/
On Tue, Feb 2, 2021 at 12:28 PM Rui Barradas <ruipbarradas at sapo.pt
<mailto:ruipbarradas at sapo.pt>> wrote:
Hello,
You get the fitted values for years 2000, ..., 2019.
Those values are the original series minus the residuals:
f <- fitted(model1)
g <- yy - resid(model1)
identical(f, g)? ? ? ? ? # returns TRUE
If you want to *forecast*, this will give you the default h = 10
forecasts.
fc <- forecast(model1)
plot(fc)
Hope this helps,
Rui Barradas
?s 23:31 de 01/02/21, Md. Moyazzem Hossain escreveu:
> Dear Rui Barradas
>
> Thank you very much for your reply.
>
> However, still now, I have a confusion whether I get the fitted
> for the year 2000, 2001, ..., 2020 or 2001, 2002, ..., 2021.
>
> Need any more help.
>
> Thanks in advance.
>
> Md
>
> On Thu, Jan 28, 2021 at 4:47 PM Rui Barradas
<ruipbarradas at sapo.pt <mailto:ruipbarradas at sapo.pt>
> <mailto:ruipbarradas at sapo.pt <mailto:ruipbarradas at sapo.pt>>> wrote:
>
>? ? ?Hello,
>
>? ? ? ?From help('forecast::fitted.Arima'):
>
>? ? ?h? ? ? ?The number of steps to forecast ahead.
>
>
>? ? ?So you have the default h = 1 step ahead forecast for your model.
>
>
>? ? ?Hope this helps,
>
>? ? ?Rui Barradas
>
>? ? ??s 12:13 de 28/01/21, Md. Moyazzem Hossain escreveu:
>? ? ? >? ?Dear R-experts,
>? ? ? >
>? ? ? > I hope that all of you are doing well. I got the filled value
>? ? ?from the
>? ? ? > ARIMA model.
>? ? ? >
>? ? ? > I use the following working code. But I am not clear whether I
>? ? ?got the
>? ? ? > fitted value for each *corresponding time* of the original
>? ? ?point like
>? ? ? > 2000, 2001, 2020 or get a *one-step-ahead* fitted value.
>? ? ?suggest me
>? ? ? > any reference for further reading to my understanding.
>? ? ? >
>? ? ? > ########################
>? ? ? >
>
?y<-c(120,340,250,430,125,324,763,458,763,905,765,456,234,345,654,654,567,876,907,456)
>? ? ? > library(forecast)
>? ? ? > library(tseries)
>? ? ? > yy=ts(y, start=c(2000,1))
>? ? ? >
>? ? ? > model1=Arima(yy,order=c(0,2,1), lambda = NULL,method='ML')
>? ? ? > model1
>? ? ? >
>? ? ? > f <- fitted( model1)
>? ? ? > plot(yy)
>? ? ? > plot(f)
>? ? ? >
>? ? ? > Thanks in advance.
>? ? ? >
>
>
>
> --
> Best Regards,
> Md. Moyazzem Hossain
> Associate Professor
> Department of Statistics
> Jahangirnagar University
> Savar, Dhaka-1342
> Bangladesh
> Website: http://www.juniv.edu/teachers/hossainmm
> Research: *Google Scholar
>