dictive accuracy. Quinlan's ID3, an early decision
tree learner, initially used the information gain split
method. But Quinlan discovered that information
gain showed unfair favoritism toward attributes with
many outcomes. Consequently, gain ratio later became
the default split method.