Creating a Response Surface Design - RSM http://www.jmp.com/support/help/Creating_a_Response_Surface_Design.shtml
JMP has a powerful Design of Experiments or simply DOE platform that can be used for almost any experimental situation including factor screening and optimizations. It's called Custom Design.
The Custom Designer can accommodate designs with hard to change factors and other constraints.
To begin go to the DOE menu and select DOE, then select Custom Design.
If you don’t have a DOE menu visible, go under Preferences, scroll down to Menu, then turn on DOE.
You can see from the DOE menu, JMP has other DOE platforms and Classical DOE designs ready when you need them, but the Custom Designer can handle most situations.
Selecting, Custom Design, the first window opens, and it's under Responses where you specify the response variables and their characteristics.
To change the default name, double-click the response name.
The response is a result of the experiment. Something you care about, that you can measure well. Like time, or weight, or yield.
Each response has a goal which will be used during an optimization phase of the analysis. To change the goal double-click the default goal, which is maximized, and change it to match target, minimize or no goal.
To add other responses click Add response.
Under factors is where you specify the experimental factors of your design. These are the things you will vary to determine their impact on the response.
Select Add factor and select the type of factor and optionally the number of levels.
Notice that the Custom Designer in JMP can handle many different types of factors.
Double-click a factor name or the factor values to change them.
The experimental settings under the “Changes” heading are where you can set the difficulty of modifying the factor levels, easy is the default.
Setting a factor to hard or very hard to change will result in a split or split-split plot design.
Click continue to proceed the next step of the design process, specifying the model to be estimated.
To add all interactions or power terms up to a given degree click the corresponding buttons.
To add terms needed to perform a response surface analysis select RSM.
You’ll find several links below to learn more about Response Surface Models.
To remove a term highlight it and click remove term.
Under design generation, we can fine-tune the design as needed.
Next, we can select or specify the total number of runs we desire.
One of the great benefits of a custom design is you can get exactly the number of runs you can afford. You’re able to maximize the resulting amount of information for the budget.
Alternative designs force you to compromise here, either spending more than you want or getting less information than you could.
Click Make Design and the generated experimental design result is displayed.
The design evaluation panel houses a variety of Diagnostics to help you compare potential designs.
Click Make Table to generate the data table for the design.
Now, run the Experiment, collect the data, and enter it into this data table.
Attached to the data table is a Model script which will automatically run the appropriate analysis for the design.
This is a very brief introduction to using Custom Design in JMP and there are resources below to learn more. There's also an upcoming video series that introduces JMP Custom Design with a variety example and best practices.
Thanks for watching.
