 Biased toward attributes with more values
 Cannot directly handle continuous data
 Sensitive to noisy data
 Can overfit the dataset