“决策树的目的(多路)算子" is?

AP

“决策树的目的(多路)算子" is? The help just emphasizes "The Decision Tree (Multiway) operator is a nested operator i.e. it has a subprocess. The subprocess must have a Tree learner i.e. an operator that expects an ExampleSet and generates a Tree model. ",what is the unique function of this operator ?

in addition，the result of its attached tutorial process is empty

jwpfau

Hi,

you have to select an element on the left.



The Decision Teree (Multiway) is a grouped model that additionally applies discretization based on the results of it's inner tree model.

Greetings,
Jonas

AP

what is the meaning of “grouped model”？

AP

What is the necessity of introducing this operator?

AP

In the tutorial process，When you use “Decision Tree (Multiway)”，You will get as below，But when you not use “Decision Tree (Multiway)”，You will get just the same，What's the necessity of “Decision Tree (Multiway)”？

jwpfau

Hi AP,

the main difference is the multisplit vs the binary tree, for the motivation of the algorithm i can only citethis paper.

...it dramatically reduces tree complexity without hurting classifier accuracy at a reasonable cost. Moreover, knowledge workers (the executives, analysts, and managers who employ decision support systems) usually feel more comfortable with multi-way splits than with binary trees, and this is also true for numerical attributes.

Greeting,
Jonas

AP

But when you use operator“Decision Tree” alone and set its parameter “criterion” to“gain_ratio”(Just as it is in the operator "Decision Tree (Multiway)"),you will the same multiway tree

jwpfau

Hi AP,

It may look the same, but one is a binary tree the other one could have multiple categorical splits.

Greetings,
Jonas