CART (Classification and Regression Trees) is a decision tree algorithm that constructs trees
for both classification and regression problems.
Unlike ID3, CART always creates binary splits, meaning each decision node has exactly two
branches.
Key Idea Behind CART
CART works by repeatedly asking a yes/no question that best separates the data.
At each step, it chooses the split that produces the purest possible child nodes.
Splitting Criteria in CART
CART uses different criteria depending on the type of problem:
 Gini Index → for classification
 Mean Squared Error (MSE) → for regression
Important Characteristics of CART
 Always produces binary trees
 Works for both categorical and continuous features
 Can handle classification and regression
 Simple and easy to interpret
 Used in many real-world applications