# Does new MMA9 Time Series support obsolete the old TimeSeries Package?

MMA9 includes a fair number of functions and symbols that seem to overlap with the old TimeSeries package. The new features are documented here and the old material is documented here. Some of the names are 'same-only-different:' for example, the old package has ARMAModel and the new built-in has ARMAProcess. The argument list and meaning appear to be the same.

I'd like to know if the old package is obsoleted by the new features, or is there still enough utility in the old package to justify its price (around \$300)? I can't find a try-before-buy option on the package, so this question seems difficult to answer on my own.

-
I've never seen a try-before-buy option on any of the commercially available packages, so I believe that's not usually done. What features do you miss in the built-in timeseries implementation? – Sjoerd C. de Vries Feb 6 '13 at 22:44
The question ~"what features do you miss in the built-in"~ is exactly what I'm trying to find an answer for :) What's in the old package that isn't in the new built-ins? Looks like there's a partial answer below: Kalman filtering. – Reb.Cabin Feb 6 '13 at 23:55
Related to the OP I was also wondering if one can fit GARCH family of models in MMA9. – Mikael Anderson Feb 7 '13 at 2:42

The package loads with a few errors eg.

SetDelayed::write: Tag CovarianceFunction in CovarianceFunction[ARModel[phi_,sigma2_],n_] is Protected. >>

SetDelayed::write: Tag CovarianceFunction in CovarianceFunction[MAModel[theta_,sigma2_],n_] is Protected. >>

SetDelayed::write: Tag CovarianceFunction in CovarianceFunction[ARMAModel[phi_,theta_,sigma2_],n_Integer?NonNegative] is Protected. >>

General::stop: Further output of SetDelayed::write will be suppressed during this calculation. >>


Some of the examples (eg BestLinearPredictor) appear to work. Some (eg. the Kalman filter material) do not.

-