What is the role of greedy optimization in Decision Trees?

Decision Trees use greedy optimization by selecting the best split at each node without considering global optimality.

The tree-building process is greedy because it chooses the best local split based on impurity reduction at each step. It does not look ahead to future splits, which means it may miss globally optimal tree structures. This is a key reason why decision tree training is fast but not guaranteed optimal. The greedy nature contributes to instability and high variance.