该帮助的版本不再更新，请参见https://www.jmp.com/support/help/zh-cn/15.2 获取最新的版本.


Scripting Guide • Scripting Platforms • Scripting Considerations by Platform • DOE
上一个
 • 
下一个
DOE
提示：For a complete list of DOE scripting options, select Help > Scripting Index, select Objects from the menu, and search for DOE.
To reproduce your DOE work, most DOE platforms provide these approaches:
•
In the red triangle menu for the DOE window, the Save Script to Script Window option provides a script that reproduces the work that you have entered. This option is not available for the Nonlinear Design and Taguchi Array platforms.
•
In the output design tables created by most DOE platforms, a script called DOE Dialog reproduces the work in the DOE window that created the design table. This script also contains the random seed (and number of starts or column starts, where appropriate) that enables you to reproduce your design exactly.
For many DOE platforms, the design table that is created also contains one or more scripts that you can run to conduct an appropriate analysis. These scripts are prefilled with the appropriate settings for your design.
The Random Seed
In constructing designs using the DOE platform, JMP uses a random seed to control certain actions that have a random component. A random seed can control the following:
•
simulating responses using the Simulate Responses option
•
randomizing Run Order for design construction
•
selecting a starting design for designs based on random starts
To reproduce a design or simulated responses using a script, you need to specify the random seed that generated them. For designs using random starts, set a random seed in the script before making the design. To control simulated responses or run order, set a random seed in the script before making the design table.
The following example creates a Custom Design, sets the random seed, and makes the design:
DOE(
Custom Design,
Add Factor( Continuous, -1, 1, "X1", 0 ),
Add Factor( Continuous, -1, 1, "X2", 0 ),
Set Random Seed( 34067086 ),
Make Design
);
Reproduce a Custom Design or a Main Effects Screening Design
Custom designs and main effects screening designs (constructed in the Screening Design platform) are generated by allocating a maximum number of seconds (Design Search Time) spent searching for a design. The default setting for Design Search Time is based on the complexity of the design. Even when two custom or main effects screening designs are constructed using the same random seed, the designs might not be identical because of different machine processing capabilities.
When you save the script for either of these designs using the Save Script to Script Window option, or when the DOE Dialog script is saved to the data table, the script specifies the Random Seed and the Number of Starts. The Number of Starts is the number of random starts used during the Design Search Time. The combination of the Set Random Seed and the Number of Starts options enables you to reproduce the design.
注意：If you write your own script, be aware that the order in which you specify options in your script can affect your results. For example, you should specify the design options before you make the design. In particular, if your design is constructed using a random seed, set it before you make the design.
Global Variables
Here are some global variables used to initiate or tune search algorithms.
Starting Design
Enables you to specify a starting design. The following example replaces the random starting design with a specified matrix:
DOE Starting Design = matrix;
If a starting design is supplied, the Custom Design platform has only one start using this design.
K Exchange Value
The coordinate exchange algorithm considers every row of factor settings for possible replacement in every iteration. You can limit the number of rows considered for exchange in an iteration so that only a small number of most likely rows is considered. The following example tells the algorithm to consider only the three most likely rows for exchange in each iteration:
DOE K Exchange Value = 3;
Bayes Diagonal
Defines a vector that is added to the diagonal elements of the X'X matrix. This new matrix is used in finding the D-optimal design. The following example adds the elements of vector to the diagonal elements of the X‘X matrix:
Bayes Diagonal = vector: