Hello,
Am trying to set up ARIMA model for the attached graph, can you please advice me with best setting I can use
is it ARIMA (1,0,0) or any other option?
Thanks
hello, from the picture you posted it looks like your series shows both a seasonal pattern and a trend. I suggest modeling seasonality with MA components: I cannot tell the order without specific information about the series; e.g. for a monthly series with a quarterly seasonal component you may use
y(t) = b*e(t-4) + e(t)
where e() is a white-noise.
the trend can be modeled either with a deterministic component, e.g.
y(t) = b*t + e(t)
or stochastically, with an AR component. in this case, check for unit roots first and, if positive, take appropriate differences of the series to make it stationary. FP
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs