The ability of JMP Software to easily create decision trees makes it a fantastic tool for troubleshooting.
I’m going to share with you 3 tips to increase their effectiveness.
Convert a continuous response into a binary response with values representing good and bad outcomes.
If this condition is relative to a specification limit then the binary response can be generated using a column formula. But more generally, the column can be created by the selecting data rows of interest and then choosing Name Selection In Column from the Row Selection submenu of the Rows menu.
For a decision based on a binary outcome each node of the tree will show a barchart representing the relative proportions of the outcomes. To quantify these proportions, and to have clearly labelled outcomes, select Show Split Prob from the Display Options red triangle submenu.
If the default colours assigned by JMP are confusing (e.g. red for good, blue for bad) this can be a source of cognitive dissonance. Use the Value Colors property of the binary response column to make more logical colour assignments.
The video below explores these ideas in more detail.