In a splitting rule at internal nodes of the tree based on thresholding the value of a single feature, it follows that a tree with k leaves can shatter a set of k instances.

In a splitting rule at internal nodes of the tree based on thresholding the value of a single feature, it follows that a tree with k leaves can shatter a set of k instances. Correct Answer True

Here the splitting rule at internal nodes of the tree is based on thresholding the value of a single feature; it follows that a tree with k leaves can shatter a set of k instances. Hence, if we allow decision trees of arbitrary size, we obtain a hypothesis class of infinite VC dimension and this approach can easily lead to overfitting.

Related Questions