Decision Tree Classifiers (DCT)
• The classifier has a tree structure where each node is either:
• A leaf node with a class label, or
• An internal node indicating some test to be carried out on the example
passing through it
• Most tree classifiers use a single attribute-based test at internal
nodes
• These classifiers are intuitive and easy to understand
Building Decision Trees aka DT Induction
A two-phase procedure consisting of growing and pruning
• The growth phase searches for successive splits to divide training
examples into smaller subsets of increasing purity or homogeneity in
a top-down greedy manner
• Splitting is generally done by finding an attribute-value pair that
maximizes some purity criterion