Decorrelating the decision trees

Question

It says in the lesson "It's important to note that the random selection of features happens at each node". Let's say: The whole dataset totally contains 9 features, so I consider 3 features (out of the 9) to be used in one node. But finally only 1 feature can be really applied on one node, correct? So my question is: Does it really make a difference between: I randomly choose 1 feature (to be used on a node) out of the 3-features subset and out of the 9-features whole-set? The 3-features subset is anyway a random choose from the 9-feature whole-set.

Answer 1

Pls show me what programing language you are talking about so that i will understand it better