[6] Information gain and impurity of decision tree

Decision trees are named because they are like trees in the form of class classification through certain criteria. The criteria for classifying decision trees are information gain. Information gains can be determined based on impurity. As the name suggests, impurity is an indicator of how various classes are mixed into the node. \( IG(D_p,f) = I(D_p) - \sum_{j=1}^{m}\frac{N_j}{N_p}I(D_j) \) Info..

2021.01.06