Reduces tree size to avoid overfitting and improve interpretability

Reduces tree size to avoid overfitting and improve interpretability

• A. Boosting
• B. Regularization
• C. Bagging
• D. Pruning

Answer: (D)

Pruning trims the tree after it’s grown, removing branches that don’t add meaningful predictive power. This reduces model complexity, which helps prevent the tree from fitting noise in the training data (overfitting) and makes the resulting decision paths easier to follow and interpret. By cutting away parts that aren’t helping, you keep the essential structure that generalizes better to unseen data. Boosting builds multiple trees sequentially to boost accuracy, not by shrinking an individual tree. Bagging also relies on creating many trees in parallel and combining their outputs. Regularization is a broader idea to penalize complexity, which can influence how a tree grows but doesn’t inherently involve post-growth reduction of the tree’s size like pruning does.

Pruning trims the tree after it’s grown, removing branches that don’t add meaningful predictive power. This reduces model complexity, which helps prevent the tree from fitting noise in the training data (overfitting) and makes the resulting decision paths easier to follow and interpret. By cutting away parts that aren’t helping, you keep the essential structure that generalizes better to unseen data.

Boosting builds multiple trees sequentially to boost accuracy, not by shrinking an individual tree. Bagging also relies on creating many trees in parallel and combining their outputs. Regularization is a broader idea to penalize complexity, which can influence how a tree grows but doesn’t inherently involve post-growth reduction of the tree’s size like pruning does.