🚀 go-pugleaf

Hello friends,

    I have some queries regarding the variance stability test
    that is done in AUTOBox
    (1) AUTOBOX selects the final model form and then
        performs variance stability test on the residuals
        from this final model.
        Does the model remain optimal after the series has 
        been treated for variance stability ? If it does,
        then can it be proved mathematically or even explained
         otherwise.
    (2)can anyone explain with examples what is the difference
        between lambda transformation for making the series stationary
        and variance stability test for making the series homogeneous.
    (3)The same person working on the same series at two different times
       may arrive at two different ARIMA models. Then , how can a structural
       change in the series be identified by looking at the model forms alone.
       Can some kind of equivalence be arrived at between two ARIMA
       models for the same series?.
-- 
Mrs. Rama     EMAIL-modeling@ern.doe.ernet.in
Analytics & Modelling div.
NIC,A-Block, CGO COMPLEX,N.Delhi,India.
.