about partition

[#1]

When distinguishing between A and B using continuous variables from tests a, b, c, and d, we employ partitioning to evaluate the contribution of these four tests. However, factors like (a) and (b) appear repeatedly in multiple splits, complicating the analysis. How can we adjust the settings to prevent specific factors from being selected too frequently?　Attach the actual results and an image of the ideal results.　Please answer the question. Thank you.

[#2]

Factors (a) and (b) show up repeatedly because they are the best factors to use (explain the most variation in the target variable). Why would you want to override that? How did you arrive at your "ideal" tree?

The same factors are likely being chosen because you have more of a linear relationship between the factors and the target variable. Many splits are required to approximate that linear relationship.

There is no automatic way to tell JMP to use a variable for splitting only once, which is what you seem to want. In fact, that idea goes contrary to how you build a tree. But, JMP does allow you to build a tree manually, where you decide which variables to split on and where to split them. 

When you get into the platform, click on the red arrow of the leaf where you want to add a split.  If you choose Split Specific you can specify the factor and the level of the input. In this fashion you can build whatever tree you want. Just remember, your tree will no longer be the best fitting tree if you override the default algorithm.