JMP DOE help on prediction profiler

[#1]

Hi all,

I am new to JMP DOE. I have a problem.

I have made my own set of runs without using JMP DOE tools. But I need to get the prediction profiler from that set. How to?

Please help, I cannot redo my experiments.

Thanks.

[#2 · Accepted solution]

Don't worry - So long as you have the results you can make some progress (though how much will depend on the design of your runs). In this case, the workflow in JMP requires that you fit some kind of model to your data (if you had generated a design via JMP you would have a saved script in the table to get you started with this step). Once you have a useful fit, the Prediction Profiler will be available in the report window. You can also elect to save the prediction formula(s) from the fit to the data table, then use the Prediction Profiler outside of the fitting platform.

Probably you need to fit some kind of linear model, so I would look at this aspect of Fit Model to start with.

[#3]

Documentation on fitting models is readily available from the pull down menus

     Help==>Books==>Fitting Linear Models

The excert below, from the book's Table of Contents, you can see there are lots of examples on how to setup the type of Fit Model you may need to use

Model Specification Specify Linear Models                       31 
	Overview of the Fit Model Platform                      32 
	Example of a Regression Analysis Using Fit Model        33 
	Launch the Fit Model Platform                           36                               
	Fit Model Launch Window Elements                        37 
	Construct Model Effects                                 39 
	Fitting Personalities                                   45 
	Model Specification Options                             47 
	Informative Missing                                     49 
	Validity Checks                                         50 
	Examples of Model Specifications and Their Model Fits   50 
	Simple Linear Regression                                50 
	Polynomial in X to Degree k                             51 
	Polynomial in X and Z to Degree k                       52 
	Multiple Linear Regression                              52 
	One-Way Analysis of Variance                            53
Two-Way Analysis of Variance                                54 
Two-Way Analysis of Variance with Interaction               55 
Three-Way Full Factorial                                    57 
Analysis of Covariance, Equal Slopes                        59 
Analysis of Covariance, Unequal Slopes                      60 
Two-Factor Nested Random Effects Model                      62 
Three-Factor Fully Nested Random Effects Model              64 
Simple Split Plot or Repeated Measures Model                65 
Two-Factor Response Surface Model                           66 
Knotted Spline Effect                                       68 

[#4]

Thank you very much for this. I don't know much about JMP yet really and this gave me additional knowledge.

I tried running my design in 'Fit Model' and then run the profiler and got what I need. Again, thanks.

[#5]

To add to my colleague @ian_jmp and @txnelson's comments, going forward using JMP for empirical modeling purposes, the JMP Prediction Profiler permeates many, many modeling platform methods within JMP. For example, a Prediction Profiler can be found in the tree based modeling methods in the Predictive Modeling platform or the modeling based subplatforms in the Consumer Research section as well.