Enhanced Data Manipulation

Simulation Studio 1.6 enhances its ability to work with source data. The new Observation Source block enables you to sample an entire observation (or row) from a source SAS data set or JMP table in a single step; this is an expansion of the ability of the "SAS Data Column" choice for the Numeric Source block, which samples one variable at a time. This enhanced sampling capability is especially useful with models in which a great deal of data must be sampled from the same data source at one time, making such models far more compact than in the past. For example, the Observation Source block enables you to read an entire row from a data set and assign it as an entity attribute. The new dot (.) operator available in the Formula block can be used to access the values of the observation’s member variables.

Integration with JMP distribution-fitting capabilities is now incorporated into the Numeric Source block and is tighter in Simulation Studio 1.6 than in past releases. This integration enables you to use the JMP "fit all" capability to view numerous candidate distributions and graphs of their respective fits of the specified data. Selecting your choice of distribution among these candidates automatically populates the appropriate Numeric Source block with the chosen distribution and its parameter values.

Simulation Studio 1.6 also adds new capabilities for sampling from data-driven probability distributions. You can use data to specify a discrete empirical distribution (for which the data specify values and associated probabilities of occurrence) or a continuous empirical distribution (for which the data specify ordered values and corresponding cumulative probabilities). Additionally, you can specify nonhomogeneous Poisson processes (in which the arrival rate varies over time). These include count-based processes (for which the data specify time intervals and associated arrival counts) and rate-based processes (for which the data specify time intervals and associated arrival rates). More details about both empirical distributions and nonhomogeneous Poisson processes can be found in Appendix B, Random Variation in a Model.

Two new blocks, the Dataset Holder and Dataset Writer, work together to provide more flexible and more extensive access to data. The Dataset Holder block provides a repository for data and enables customized queries and extractions from the data; it enables you to view and access the entire data set and does not limit you to a single variable or a single observation. The Dataset Writer block enables you to create output data at any point during the simulation run. Collectively, the Dataset Holder and Dataset Writer blocks enable event-driven data interactions (read and write) throughout the simulation run. Each is compatible with both SAS data sets and JMP tables.

The Stats Collector block expands your ability to calculate statistics on simulation-generated data, generalizing capabilities found in the Queue Stats Collector and Server Stats Collector blocks to work with any specified sources of data. Finally, the new Stopper block enables you to create an event that immediately stops the simulation run and can also trigger the saving of key simulation data near or at the end of the simulation run.