CSSM Procedure

SIMSMOOTH Statement

SIMSMOOTH <INMODEL=blob-table> <OUTMODEL=blob-table> <options> ;

The SIMSMOOTH statement enables you to use the fitted model for simulation smoothing. (For more information, see the section Simulation Smoothing.) The OUTMODEL= option enables you to store the fitted model in a binary large object (BLOB) table to use later for simulation smoothing. The INMODEL= option enables you to load a previously stored fitted model and use it for simulation smoothing. The other options in this statement enable you to control different parts of the simulation smoothing process. For an illustration of simulation smoothing, see Example 14.21: Simulation-Smoothing-Based Sampling Distribution of the Sum of Forecasts.

You can specify the following options:

INMODEL=blob-table: specifies a previously created BLOB table that contains the fitted model for the input data table. When you specify this option in the SIMSMOOTH statement, the statements that are used for model specification (such as the MODEL, STATE, and TREND statements) are not needed. In addition to the PROC CSSM and SIMSMOOTH statements, only the options that control the simulation smoothing steps, such as the NSIM=, SEED=, and OUTPRED= options, are used.
NSIM=number: specifies the number of simulations to generate, where number is a positive integer. By default, NSIM=1.
OUTCOMP=libref.data-table: specifies an output data table to store the simulated components. This table contains the ID statement variable, the OBSINDEX statement variable (if specified), and the simulated components. For more information, see the section Output Tables Associated with Simulation Smoothing.
OUTMODEL=blob-table: specifies a BLOB table to store the fitted model specification for later use in simulation smoothing. When you specify this option in the SIMSMOOTH statement, other options in this statement (such as the INMODEL= or NSIM= option) are ignored.
OUTPRED=libref.data-table: specifies an output data table to store the simulated predictions of the missing response values. This table contains the ID statement variable, the OBSINDEX statement variable (if specified), and the simulated predictions of the missing response values. For more information, see the section Output Tables Associated with Simulation Smoothing.
OUTSTATE=libref.data-table: specifies an output data table to store the simulated latent states. This table contains the ID statement variable, the OBSINDEX statement variable (if specified), and the simulated states. For more information, see the section Output Tables Associated with Simulation Smoothing.
SEED=seed: specifies the start of the set of random number seeds to use during the simulation process. As an example, if the number of simulations that you specify in the NSIM= option is 100 and SEED=5, then the seeds between 5 and 104 are sequentially used for random number generation during the 100 simulations. By default, SEED=1.

The SIMSMOOTH statement is used both in creating the fitted-model-specification BLOB table (using the OUTMODEL= option) and later in simulation smoothing (using the INMODEL= option).

Note: Simulation smoothing is not currently supported in situations where temporal aggregation or temporal distribution is requested. For more information, see the AGGREGATE and DISTRIBUTE options in the MODEL statement.

Last updated: July 09, 2026