Algorithm 1: General learning algorithm for decision trees with two classes (A and B)

1Create a root node v for the decision tree. Let D[v] denote the learning data sample. 2If D[v] contains examples of the A class only, mark the node v as terminal, and assign a class label A. Return. 3If D[v] contains examples of the B class only, mark the node v as terminal, and assign a class label B. Return. 4Try to find the most discriminant attribute and condition on it for the data sample D[v]; otherwise, mark the node v as terminal, and assign a class label represented by most of the data sample D[v]. Return. 5Add two edges to the node v with two new nodes v1 and v2. Split the data sample D[v] into two new samples, D[v1] and D[v2], according to the condition found. 6Run the algorithm for v1 and D[v1]. 7Run the algorithm for v2 and D[v2].