56514_en-US

liliconcom

Can I capture the data used to create the shaded confidence interval area on a marginal model plots

When using the "asses variable importance" options inside profiler for linear regression you can choose to see marginal model plots (Prediction Profiler > Asses Variable Importance > Marginal Model Plots > Select Data). 

 

One of the options for visualization is to be able to see the confidence intervals for each variable (attached file). Obviously, JMP is calculating the shaded confidence interval area somehow but there is no way to capture that range as an output.  I would like that data.

 

Is it possible to recreate that data manually, or to script a solution, or something I'm not thinking of so that I can get it as an output?       

 

Re: Can I capture the data used to create the shaded confidence interval area on a marginal model pl

This is great, thank you!  

 

I haven't had a chance to try any of this out yet and I'm going to be out of office through 5/22 so I'm going to leave this post unresolved for the time being.  

 

This is mostly to let community admins know that I will address the post again in a few weeks.  We'll also be able to see if anyone else has anything else to add to it.  

Re: Can I capture the data used to create the shaded confidence interval area on a marginal model pl

Many JMP analyze platforms have messages to save information like confidence intervals, and many have JSL only statements to get more.

 

I did not find anything about getting the Marginal Model Plots Confidence intervals.

 

There is a method to obtain what you see, but it requires some JSL know-how. For any FrameBox, you can use  the message Get Journal which returns a string that is a mix if JSL like statements. This will contain the values if the confidence band which is a PolySeg feature.   See the attached script.

 

Usng XPath the messages to change the colors or fill  or line type for a PolySeg can be done, but it seems that the coordinate values are not exposed. [I am interested if this is doable.]  

 

If the Marginal Plot was not created with a script, you can get a handle to it.