Description

The Numeric Source block provides a source of random variation using pseudo-random number generators. This block can also read numbers from a SAS data set or JMP data table.

A collection of discrete and continuous distributions are available, or you can provide the file path for the SAS data set or JMP data table along with the numeric variable column name.

The data provided by the Numeric Source block can be viewed as a stream of numbers, and the numbers are pulled from the stream one after another during a simulation. For example, each time a value is pulled from the OutValue port of a Numeric Source block, the block outputs a number from its current data stream by generating a sample from its related distribution or by reading a value from the data set, whichever is appropriate. If the last observation is reached when reading from a data set, the process resets to the beginning of the column.

The Numeric Source block provides three types of data streams: Theoretical, Fitted, and Data Driven. The Theoretical data streams include a collection of theoretical discrete and continuous distributions. The Fitted data streams are these same distributions that can be fitted using an input data variable (column) in a SAS data set or JMP data table. After the parameters of the distributions are estimated, the input data set is not needed during a simulation run. The Data Driven data streams require an input data set and include options such as empirical distributions and nonhomogeneous Poisson processes, in addition to the numeric SAS data column.

Fixed Ports

InUpdate

Input Boolean value port that signals an update of input data and stream parameter specifications. The new specifications are pulled from the InStreamPolicy or InDataPolicy ports (or both) if these ports are connected. A false Boolean value is ignored.

InStreamPolicy

Input string value port that allows new stream parameter specifications to come in after an update signal is received.

OutValue

Output number port for the numeric values to be pulled.

Dynamic Ports

InDataPolicy

Input string value port that allows new input data specifications to come in after an update signal is received. This port is available only for the Data Driven data streams.

Dialog Box Controls

Specify the type of the data stream to be provided by this Numeric Source block.

When the Theoretical type is selected, the following dialog box controls are enabled:

Type

From the list, select the distribution to sample from. The Parameters section will be updated to reflect the distribution.

Stream Parameters

This area provides fields for modifying the parameter values associated with the selected distribution. Each distribution has a Random Stream Seed entry field. Although Simulation Studio automatically assigns a different seed for each source of randomness, you can use this field to specify the seed value. Valid values for this field are integers in the range of 0 to the Java Long.MAX_VALUE.

When the Fitted option is selected, the following dialog box controls are enabled:

Input Data

This area provides fields to specify the file path of input SAS data set or JMP data table and the variable (column) name for distribution fitting. Click Fit Distribution to send the fitting request to the JMP program to perform the fitting. If you do not provide the file path or variable name, you can provide these directly using the JMP user interface before the fitting operation proceeds. The fitting results can be sent back from the JMP program to update the rest of the dialog box contents, including the associated distribution Type and Stream Parameters controls described below. You can also use these controls to adjust the parameters.

Type

From the list, select the distribution to sample from. The Parameters section will be updated to reflect the distribution.

Stream Parameters

This area provides fields for displaying and modifying the parameter values associated with the selected distribution. Each distribution has a Random Stream Seed entry field. Although Simulation Studio automatically assigns a different seed for each source of randomness, you can use this field to specify the seed value. Valid values for this field are integers in the range of 0 to the Java Long.MAX_VALUE.

When the Data Driven option is selected, the following dialog box controls are enabled:

Type

From the list, select the distribution or data stream to sample from. The following Input Data and Stream Parameters section will be updated to reflect the selection.

Input Data

This area provides fields to specify the file path of an input SAS data set or JMP data table and the variable or column names. The Load from Remote SAS Workspace Server checkbox indicates that the input SAS data set file is to be loaded from the remote SAS Workspace Server host specified in the Configuration dialog box.

Stream Parameters

This area provides fields for modifying the parameter values that are associated with the selected distribution or data stream. Each distribution has a Random Stream Seed entry field. Although Simulation Studio automatically assigns a different seed for each source of randomness, you can use this field to specify the seed value. Valid values for this field are integers in the range of 0 to the Java Long.MAX_VALUE.

If the selected data stream type is SAS Data Column, fields to enter Lazy Loading, and Reset At Updated are displayed. Both fields are Boolean. If the Lazy Loading field is false, the input data set file has to be loaded at the start of simulation. Otherwise, the data file is loaded only when its contents are needed during simulation. If the Reset At Updated field is true, an update signal value of true causes the first observation in the updated data set to be the observation pulled next.

For all other Data Driven stream types, the lazy loading option is also supported.

Candidates for Design of Experiments

Factors

DataStreamDescription (text)
The format for specifying the value of the DataStreamDescription factor is as follows:
class==dataStreamClass;attribute1==attribute1Value;...;attributeN==attributeNValue
where:

dataStreamClass

is the fully-qualified Java class name of a data stream type.

attribute1

is the name of the first parameter associated with the specified data stream type.

attribute1Value

is the value of the first parameter associated with the specified data stream type.

attributeN

is the name of the last parameter associated with the specified data stream type.

attribute1Value

is the value of the last parameter associated with the specified data stream type.

For example, possible values for a DataStreamDescription factor might be as follows:

class==com.sas.analytics.simulation.datastream.distribution.Exponential;Mean==4

class==com.sas.analytics.simulation.datastream.distribution.Normal;Mean==1;Std Dev==2

If you want to keep the same distribution but alter the parameter setting(s) for the distribution, you only need to supply the information needed for that parameter you want to change. For example, if you are using an Exponential distribution and you want to change the Mean value associated with the distribution, possible values for the DataStreamDescription factor might be as follows:

Mean==4

Mean== 4.5

Responses

None