CARIMA Procedure
ESTIMATE Statement
ESTIMATE <options>;
The ESTIMATE statement specifies the type of ARIMA model to be fit for the variable in the preceding IDENTIFY statement.
You can specify the following options:
- CONVERGE=value
specifies the convergence criterion, where value is a number between 0 and 1, exclusive. Convergence is assumed when the largest change in the estimate for any parameter is less than the specified value. If the absolute value of the parameter estimate is greater than 0.01, the relative change is used; otherwise, the absolute change in the estimate is used. By default, CONVERGE=0.001.
- DELTA=value
specifies the perturbation value for computing numerical derivatives, where value is a number between 0 and 1, exclusive. By default, DELTA=0.001.
- DIFF=(value-list)
specifies the degrees of differencing for the variable-list that is specified in the preceding IDENTIFY statement. For example, you can specify DIFF=(1) to take a first-order difference for all variables specified in the variable-list of the preceding IDENTIFY statement. If you have monthly data and expect seasonal patterns, you can specify DIFF=(1,12) to difference all variables in the variable-list of the preceding IDENTIFY statement twice: once at lag 1, and then again at lag 12. Each analyzed series is then the difference between the current period-to-period change in the specified variable and the change 12 periods ago. Each value in the value-list is an integer greater than 0. By default, DIFF=(0).
- MAXITER=value
specifies the maximum number of iterations, where value is a positive integer. By default, MAXITER=50.
- METHOD=method-name
-
specifies the estimation method to be used.
You can specify the following forecasting method-names:
- CLS
specifies the conditional least squares method.
- ML
specifies the maximum likelihood method.
- ULS
specifies the unconditional least squares method.
By default, METHOD=CLS.
- MU=value
specifies a constant term for the ARIMA model, where value is a number. By default, MU=0.
- NOINT=0 | 1
-
specifies whether to include an intercept in the model. You can specify the following values:
- 0
includes an intercept term in the model.
- 1
omits the intercept term from the model. This suppresses the fitting of a constant (or intercept) parameter in the model. That is, the value specified in the MU value is omitted.
By default, NOINT=0.
- NOSTABLE=0 | 1
-
specifies whether to restrict the candidate estimates for the autoregressive and moving average parameters in the stationary and invertible regions, respectively. You can specify the following values:
- 0
restrict candidate estimates in stable and invertible regions.
- 1
do not restrict candidate estimates.
By default, NOSTABLE=0.
- P=value
specifies the degree of the autoregressive polynomial, where value is an integer greater or equal than 0. By default, P=0.
- Q=value
specifies the degree of the moving average polynomial, where value is an integer greater or equal than 0. By default, Q=0.
- SINGULAR=value
specifies the criterion for checking singularity, where value is a number between 0 and 1, exclusive. If a pivot of a sweep operation is less than value, the matrix is deemed singular. Sweep operations are performed on the Jacobian matrix during final estimation and on the covariance matrix when preliminary estimates are obtained. The default is 1E–7.
- TRANSFORM=transform-option
-
specifies the time series transformation to apply to the series in the ESTIMATE statement. You can specify the following transform-options:
- BOXCOX(n)
applies a Box-Cox transformation with parameter number, n, where n is a number between –5 and 5.
- LOG
applies a logarithmic transformation.
- LOGISTIC
applies a logistic transformation.
- NONE
dos not apply a transformation.
- SQRT
applies a square-root transformation.
By default, TRANSFORM=NONE.