Using Prediction Profiling to Maximize Model Proficiency - Part 1
Created:
| Last Modified:
Video was updated for JMP 18 June, 2024.
The JMP Prediction Profiler is an option available from over 20 places in JMP to use DOE, observational and formula data to visualize and explore models and make decisions.
See how to:
- Understand the purpose of Prediction Profiler, types of data that can be profiled, and how to access the Profiler
- Explore the Opportunity Space
- Terminology
- Profile trace and values shown on interactive Profiler
- Confidence Interval for the Predicted Response Mean
- Prediction Interval for finding where predicted results will be (new in JMP 18)
- Understand the model by interpreting the Profiler results
- Overlay Interactions to find potential interactions (new in JMP 18)
- Change factor levels interactively
- Use Extrapolation Control (and warnings) to see if underlying data support predictions, especially for observational data
- Turn on Data Points to see how far they are from surface of model (new in JMP 18)
- Change Profiler Appearance
- Arrange in Rows
- Remove variables
- Reorder variables
- Hide variables
- Assess and/or animate variable importance
- Share Profiler
- By publishing to File
- By saving formula to column, rerunning Profiler, then saving as Interactive html
Questions answered by Andrea @andreacoombs1 and Mike @MikeD_Anderson during the live webinar:
Q: When you hide the factors from the Profiler, are they removed from the underlying equations?
A: No. they are still being included in the interactions, they are just not being displayed in the Profiler. If you want to remove factors from a model, you can do that from Effects Summary.
Q: Can you change the variable importance animated .gifs like you do with columns switcher or local data filter in Graph Builder?
A: No. You can play it but not record it as an animation.
Q: Where is Factor Importance found?
A: Under red triangle, Assess Variable Importance. The most important thing is to choose Colorize the Profiler. You can also reorder factors by main effect or total importance.
Q: Can I change number of decimal points on the Responses?
A: You can change it in the Data Table using Column Info>Format for the Response. That changes the display for the Numeric data, but the value retains the full, longer value of the data to be used in calculations.
Q: Can you constrain factors to a certain range instead of locking them?
A: Yes, you can Add Linear Constraints. If your data is from a designed experiment where you defined the constraints for your design, the constraints from the DOE will automatically be applied.
Q: Will you please show, in general, how to save the formulas to get the confidence intervals in case you want to graph them?
A: In JMP 18 you save Prediction and Interval Formulas. In JMP 17, here is how to save Standard Error Prediction Formula and Confidence Intervals:
Resources
- Part 2 - Using Prediction Profiling to Maximize Model Efficiency video.
- Controlling Extrapolation in the Prediction Profiler in JMP Pro Discovery Summit 2021 presentation
- Prediction Profiler documentation
