I have a modelling excercise where I want to apply a decision tree model to a dataset with continuous variables. The target is a binary outcome. At each node I want division into 3 cases, pure 1; mixed 1,0; and pure 0. and the next split applied to the mixed case, in the same way. The resulting tree looks like a main trunk, with one leaf on either side of each node. Is it possible to constrain the decision tree process to achive this?
The software does not support that constraint. In fact, there may be few data sets that contain candidate splits that satisfy that constraint. You may want to use the interactive tree function to explore splits and find out if that is even possible.