cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
hogi
Level XI

Interaction Profiler: show Datapoints?

In Jmp18, there is a new feature to activate data points in Prediction Profiler enhancements in JMP® 18 

Is there an option to activate the data points as well for interaction profiler?

hogi_0-1715002218694.png

 

1 ACCEPTED SOLUTION

Accepted Solutions

Re: Interaction Profiler: show Datapoints?

Try turning on the Overlaid Interaction option.   It allows you to see the interactions in the profiler itself.  

View solution in original post

4 REPLIES 4

Re: Interaction Profiler: show Datapoints?

Try turning on the Overlaid Interaction option.   It allows you to see the interactions in the profiler itself.  

hogi
Level XI

Re: Interaction Profiler: show Datapoints?

ah, right


it helps to specify colors for the different rows. then the data points in prediction profiler get colored as well - thanks for this nice feature
like in this example where colors are specified by X2 - i.e. matching the second row of the interaction profiler.

 

Re: Interaction Profiler: show Datapoints?

also notice that if you hover over an interaction curve, it shows as a vertical tick indicator the levels of the other variables

SamGardner_0-1715029455109.png

 

hogi
Level XI

Re: Interaction Profiler: show Datapoints?

yes, I noticed.
another way to get the information:  the hover label

hogi_0-1715056668788.png


Nevertheless, it takes a while to get the "full picture" by hovering over individual interaction lines and checking the respective variable settings. Here I prefer the color coding in the interaction profiler - it allows "my eye" to process this information
(Plenary: All Graphs Are Wrong, but Some Are Useful by @XanGregg )

 

On the other hand, how to squeeze all the color coding of several interaction profiles simultaneously into the interaction lines of the prediction profiles? A neutral grey seems the only option.

 

Maybe: add some quick select buttons, to (gradient) color code the lines which belong to one variable - keeping the interaction with others variables gray. Then the user can pick the interaction which is most prominent and highlight it.

or: use different colors for interactions which belong to different variables.

(e.g. all green interaction curves are interactions with X1, all blue curves are interactions with X2, all red curves belong to X3.)