Decision trees make predictions by learning a hierarchy of if-then rules from data — one of the most intuitive ML algorithms.

How It Works

The algorithm learns which feature splits best separate the classes at each node, building a tree of decisions that ends in a predicted class or value.

Strengths

• Highly interpretable — you can read the rules
• Handles both numerical and categorical features
• No feature scaling needed
• Fast to train and predict

Weaknesses

• Prone to overfitting
• Unstable — small data changes create very different trees

In Practice

Rarely used alone — almost always used as base learners in Random Forest or Gradient Boosting ensembles.

Reference:

scikit-learn decision trees

https://scikit-learn.org/stable/modules/tree.html