Is this correct, If there is autocorrelation in a time series use a AR 2 model and if there is seasonality use a AR 1 model with seasonal adjustment. To test the both above t= autocorrelation /(1/(T)^.5) can be used
AR models - when we talk about AR1 or AR2 we’re talking about lags, meaning how many variables there are in the model.
AR1 model will be intercept + lag 1, AR 2 model will be intercept + lag 1+lag 2
what they’re trying to say is that if you have seasonality you can make a seasonal adjustment.
the test that you are referring to would be amongst the error terms of the residuals
To test for autocorrelation in AR Model we have to use a t-test, not DW
does that help?
To check for seasonality do we use the above t test
Bump
Any expert ideas