Fig. 13From: Lifelong Machine Learning and root cause analysis for large-scale cancer patient dataThe Decision Tree split process. Part of the tree is displayed here. The information gain for each feature variable is calculated. The feature with the highest information gain is chosen as the root and splitting node. In each node, information gain is shown as a value along with impurity level which is measured after each split. The process is iterated until the leaf nodes are generatedBack to article page