After the Decision Tree is constructed, each case i is assigned to each leaf j where j=1...N with a weight wij.
This weight wij is 0 or 1 if every test attribute is known for i. At the beginning, the whole population forms the root
of the tree. For generating different branches of the tree, the discriminating features are selected from the population
characteristics. These features are called tests which generate new child nodes. The power of discrimination can be
measured by Shannon entropy gain