After you select DOE > Custom Design, click the red triangle icon on the title bar to see the list of commands available to the Custom designer (Click the Red Triangle Icon to Reveal Commands). The commands found on this menu vary, depending on which DOE command you select. However, the commands to save and load responses and factors, and the command to set the random seed are available to all designers. You should examine the red triangle menu for each designer you use to determine which commands are available. If a designer has additional commands, they are described in the appropriate chapter.
Click the Red Triangle Icon to Reveal Commands
creates a data table containing a row for each response with a column called Response Name that identifies the responses. Four additional columns identify more information about the responses: Lower Limit, Upper Limit, Response Goal, and Importance.
Design Role that identifies the factor as a DOE factor and lists its type (continuous, categorical, blocking, and so on).
Factor Changes that identifies how difficult it is to change the factor level. Factor Changes options are Easy, Hard, and Very Hard.
Tip: It is possible to create a factors table by keying data into an empty table, but remember to assign each column a factor type. Do this by right-clicking the column name, selecting Column Info, and then selecting Column Properties > Design Role. Lastly, click the button in the Design Role area and select the appropriate role.
3.
Click the red triangle icon on the title bar and select Save Constraints. Save Constraints creates a data table that contains the information you enter into a constraints panel. There is a column for each constraint. Each has a column property called Constraint State that identifies it as a ‘less than’ or a ‘greater than’ constraint. There is a row for each variable and an additional row that has the inequality condition for each variable.
The window that appears shows the generating seed for that design (Setting the Random Seed). From this window, you can set a new random number and then run the design again.
Setting the Random Seed
Often, when you define a custom design (or any standard design), it may be useful to look at properties of the design with response data before you have collected data. The Simulate Responses command adds random response values to the JMP table that the custom designer creates. To use the command, select it before you click Make Table. When you click Make Table to create the design table, the Y column contains values for simulated responses.
y = 21 + 4X1 + 6X2 – 5X1X2 + random noise,
where the random noise is distributed with mean zero and standard deviation one.
Example of a Custom Design with Simulated Responses
1.
Select DOE > Custom Design.
3.
Click on Interactions > 2nd and select Save X Matrix from the drop-down menu of Custom Design.
4.
Using the Default Number of Runs (12), click Make Design and then Make Table.
5.
If it is not already open, select View > Log (Window > Log on the Macintosh).
6.
Click on the Moments Matrix red triangle in the upper left panel of the data table under Custom Design and select Run Script. The result shows in the log as N Row(::Moments):7, which is the number of rows in the global matrix called Moments. The Moments Matrix is dependent upon the model effects but is independent of the design. (The model effects can be viewed by clicking the red triangle by Model in the upper left panel of the data table and clicking on Run Script.) The Moments Matrix script for this example displays the value of each moment and is shown by clicking on the red triangle of the Moments Matrix and selecting Edit:
7.
Click on the Design Matrix red triangle in the upper left panel of the data table under Custom Design and select Run Script. The result shows in the log as N Row(::X):8, which is the number of rows in the global matrix called X. The X Matrix is dependent upon the design for the experiment. The script for this example shows the underlying design of the X matrix and is viewed by clicking on the red triangle of the Design Matrix and selecting Edit:
where M is a moments matrix of the parameter space that is independent of the design and can be computed in advance, and where f(x)' denotes a row of the design matrix corresponding to factor combinations of x. For additional details concerning moments and design matrices, see Myers, Montgomery, and Anderson-Cook (2009, pp. 365-371). Note that the moment matrix is called a matrix of region moments in this book. The design matrix is also called the model matrix in some books (Goos and Jones, 2011).
The default criterion for Recommended is D-optimal for all design types unless you have used the RSM button in the Model panel to add effects that make the model quadratic. For specific information about optimality criterion, see Technical Discussion.
To override the default number of random starts, click the red triangle icon in the Custom Design title bar (Click the Red Triangle Icon to Reveal Commands) and select Number of Starts. When you select this command, the window shown in Selecting the Number of Starts appears with an edit box for you to enter the number of random starts for the design you want to build. The number you enter overrides the default number of starts, which varies depending on the design.
Selecting the Number of Starts
Note: If the design iterations are taking too long, click the Cancel button. The Custom Designer stops and gives the best design found at that point.
If you have designed any factor’s changes as Hard (see Factors that are Easy, Hard, or Very Hard, to Change: Creating Optimal Split-Plot and Split-Split-Plot Designs, and Creating Random Block Designs), the sphere radius item becomes unavailable. Conversely, once you set the sphere radius, you cannot make a factor Hard to change.
For example, in a market research choice experiment, you might want to exclude a choice that allows all the best features of a product at the lowest price. In this case, the factor Feature has levels of worst (1), medium (2), and best (3), and the factor Price has levels of high (1), medium (2), and low (3). You want to exclude the third Feature level (best) and the third Price level (low).
2.
Click the red triangle icon in the title bar (Click the Red Triangle Icon to Reveal Commands) of the designer window and select Disallowed Combinations. Note that this menu item is not available if you have already defined linear inequality constraints.
3.
Enter a Boolean expression that identifies what you do not want allowed (Enter a Boolean Expression). JMP evaluates your expression, and when it sees it as true, it disallows the specified combination.
For example, in Enter a Boolean Expression, Feature==3 & Price==3 will not allow a run containing the best features at the lowest price. If there were two disallowed combinations in this example, you would use Feature==3 & Price==3 | Quality==3 & Price==3, which tells JMP to disallow a run with the best features at the lowest price or a run with the best quality and lowest price.
Enter a Boolean Expression
No Row Contains L3 for Both Price and Feature
1.
Select DOE > Custom Design.
2.
Click the red triangle icon in the title bar (Click the Red Triangle Icon to Reveal Commands) of the designer window and select Advanced Options > Mixture Sum.
1.
Select DOE > Custom Design.
3.
Click the red triangle icon in the title bar (Click the Red Triangle Icon to Reveal Commands) of the designer window and select Advanced Options > Split Plot Variance Ratio.
If you have specified If Possible as the Estimability for any factors in your model, then you can use this option to also specify the weight used for these terms. Default values are one. Larger values represent more prior information and a smaller variance. Variances are the squared values of the reciprocals of the entered values.
1.
Select DOE > Custom Design.
2.
Click the red triangle icon in the title bar (Click the Red Triangle Icon to Reveal Commands) of the designer window and select Advanced Options > Prior Parameter Variance.
By default, delta is set to 2. The default coefficient for each continuous effect is set to 1. An n-level categorical factor is represented by n–1 indicator variables. The default coefficients for the n–1 terms representing a categorical factor are alternating values of 1 and -1. The default coefficients for an interaction effect with more than one degree of freedom are also alternating values of 1 and -1.