Time Series Model Package
ARIMASPEC Object
The ARIMASPEC object generates autoregressive integrated moving average (ARIMA) model specifications for use with a TSM object.
Table 17.6 summarizes the methods that are associated with the TSM.ARIMASPEC method.
Table 17.6: Methods of the ARIMASPEC Object
Method | Description |
|---|---|
Add an autoregressive polynomial factor to the ARIMA model | |
Add a moving average polynomial factor to the ARIMA model | |
Add a transfer function to the ARIMA model | |
Add a transfer function denominator factors to the ARIMA model | |
Add a transfer function numerator factors to the ARIMA model | |
Close the ARIMA model specification | |
Open the ARIMA model specification | |
Add differencing to the ARIMA model | |
Specify the ARIMA option | |
Specify the transfer function transform | |
Specify transform |
The basic execution pattern for using an ARIMASPEC object follows this sequence of operations:
Declare: The object declaration statement creates a new ARIMASPEC object. At creation, its default state is ARIMA(0,1,0) with intercept (random walk with drift).
Open: The ARIMASPEC.Open method readies the default ARIMASPEC object for new configuration.
Configure: The various ARIMASPEC.Set methods configure the ARIMASPEC object.
Close: The ARIMASPEC.Close method finalizes the ARIMASPEC object.
Apply: The TSM.Initialize method adds the ARIMASPEC to a TSM object.
You can also store the XML representation of the ARIMA model in a CAS table. For more information, see the TSMSPEC object. Figure 17.3 illustrates the data flow through the ARIMASPEC object.
Figure 17.3: ARIMASPEC Object Data Flow
You can include series transforms such as log or Box-Cox in the specification. Deferred seasonality wildcards for seasonal ARIMA polynomial factors and seasonal differencing lags are also supported. The following list summarizes where these features can be used. For more information, see the method descriptions.
ARIMASPEC.SetDiff supports a seasonal wildcard lag in the DiffArray.
ARIMASPEC.AddARPoly and ARIMASPEC.AddMAPoly support addition of seasonal ARIMA polynomial factors.
ARIMASPEC.AddTF supports a seasonal wildcard lag in the DiffArray.
ARIMASPEC.AddTFNumPoly and ARIMASPEC.AddTFDenPoly support addition of seasonal ARIMA polynomial factors for the transfer function numerator and denominator, respectively.
ARIMASPEC.SetTransform applies transforms such as log, square root, logistic, or Box-Cox to the model.
When you specify ARIMA model coefficient values for any model components, you must specify them for all model components without regard to the value of the NOEST argument in the ARIMASPEC.SetOption method. Failing to specify a complete set of ARIMA model parameters results in an error when you call ARIMASPEC.Close, and the accumulated ARIMA model specification is reset.