During a tree's induction using the ID3 algorithm, assume the root note of the tree is at level 0. This root node represents a dataset (D) of size 10. The root node's direct child nodes are at level 1 and they represent datasets (D1 and D2) of sizes 5 and 5, respectively. D3 is a dataset of size 3 that is represented by a direct child of the node that represents D2. In D3, 1 and 2 instances are of target class A and B, respectively. The ID3 algorithm is provided with D3 and d (the current set of descriptive features) as input. If the maximum permissible depth of the tree has not been reached and the ID3 algorithm returns a leaf node with a class label of B, how many descriptive features are used to represent the instances in D?

Regression trees can use Gini index as a measure of homogeneity, such that a Gini Index of 0 represents complete homogeneity.

When inducing a classification tree using the information gain metric, the feature that yields the largest reduction in node impurity is selected for partitioning the dataset.

If a dataset (D) has equal numbers of class A, class B and class C instances and a split on the dataset results in 3 partitions each with equal numbers of only two of the classes and no instances of the third class, what is the information gain based on entropy?

Which of the following statements is true about decision tree pruning?

In general, oblique trees result in less overfitted trees than non-oblique trees, when both trees are trained on the same training set?

In general model trees are less sensitive to outliers in the training set than ordinary regression trees, when both trees are trained on the same training set.