First, let's learn more...
As Aashiq Reza brought the description link, I think the ACF
and PACF
plots that you shared is like an MA(2)
process. ARIMA(p, i,q)
has three elements, p
is for AR
, i
is for difference, and q
stands for MA
process lag. Because the lag parameter defines the lag in the model's regression formula, if both of p
and q
be zero, then the model is not ARIMA
anymore.
My suggestion: probabilistic model selection...
You can evaluate the correctness of a model for a time-series object using information criteria
like AIC
and BIC
. For example, you have a preset of possible p
and q
, then you can test each one and get the criteria for it. The model with the least criterion is the best one. This link helps with the calculation in python.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.