Developer Tutorial: Using JMP Pro Generalized Regression to Analyze Designed Experiments

[#1]

DOE allows multiple input factors to be manipulated, determining their effect on a desired output (response). By manipulating multiple inputs at the same time, DOE can identify important interactions that may be missed when experimenting with one factor at a time.

Generalized Regression used in conjunction with a designed experiment may provide answers to questions such as:

• What are the key factors in a process? (Explanatory)
• At what settings would the process deliver acceptable performance? (Predictive)
• What are the key, main, and interaction effects in the process? (Explanatory)

Variable Selection is used to:

• Fit a sequence of models (maybe many models)
• Use some metric to see how well each model fits our data
• Keep the model that fits best

In this video, see how to:

• Compare different models
• Decide on the sequence of models to fit
• Apply technique to some examples

Q&A is included throughout the video.

Resources

• Generalized Regression overview 
• Generalized Regression estimation methods, including Best Subset, Pruned Selection, Forward Selection, Dantzig Selector
• Effect Heredity documentation -   effect heredity relates to the inclusion in the model of lower-order components of higher-order effects
• Curve DOE Analysis documentation - available only if a grouping variable and at least one supplementary variable are specified in the launch window
• JMP Wish List for submitting suggestions on software enhancements that would be useful to your organization

Continue the dialog in the comments below.  Direct any questions to Clay Barker @clay_barker.