Use divide and conquer as with most tree algorithms. Create branch for each attribute value. partition the instances based on those values. Stop when Root have the same class label. ![[Screenshot 2023-07-13 at 11.05.21 am.png]]![[Screenshot 2023-07-13 at 11.09.34 am.png]] We pick the attribute to partition the root node instance based on [[entropy]] We want to get the smallest tree ![[Screenshot 2023-07-13 at 11.31.51 am.png]] ![[Screenshot 2023-07-13 at 11.32.12 am.png]]