Decision Trees (DT)

• breaks input space into regions and has separate parameters for each region
• is a boolean-valued function
• is a type of supervised learning algorithm
• can be either:
  • parametric - if decision tree algorithm is regularized with size constraints
  • non-parametric - if decision tree algorithm is allowed to learn a tree of arbitrary size

Ways to Stop Over-Splitting

• maximum depth
• minimum records per node

Look-Ahead

• DT algorithms uses greedy approach which may not always make the best splits down the tree
• look-ahead helps, but is computationally intensive

Subpages

• Classification and Regression Tree (CART) - Regression Tree
• Iterative Dichotomiser 3 (ID3)

Resources

• https://explained.ai/decision-tree-viz/index.html