What is structural bias in Decision Trees?

Structural bias refers to limitations introduced by axis-aligned splits and greedy construction in decision trees.

Structural bias arises because decision trees are constrained to axis-aligned splits and built greedily. This restricts the hypothesis space, making it difficult to represent diagonal or smooth decision boundaries efficiently. Even if data suggests a complex nonlinear relationship, trees approximate it with many stepwise partitions. This introduces bias in model representation. Oblique trees and ensembles reduce this limitation by increasing representational flexibility.