perform series of binary questions/query at each node
Partition the feature space to minimize the impurity of the partitions. A key advantage of the recursive binary tree is its interpretability.
- Non-linear model
- Can be used for classification and regression
Properties
- Interpretable / generated rules
- Categorical features are supported natively
- High-variance
Parameter
- the
max_depthof the trees controls the degree of feature interactions
- the
min_sample_leafcontrols the number of sample per leaf
Decision tree Notion
Classification
- Cost-complexity pruning
- Max leaf nodes search
