Keratoconus Severity Classification Using Features Selection and Machine Learning Algorithms
Algorithm 4
CART algorithm.
Input: Sample S
Start: Initialize the current tree to the empty tree; the root is the current node.
Repeat
Calculate the Gini(p) index of the current node p
If the current node p is terminal, then
Assign it a class.
Else
Select a test and create as many new child nodes as there are possible answers to the test: Choose the test t that maximize ∆(p, t). Where p is the current position, t is a test and Pg and Pd are the proportions of elements on the positions p1and p2, respectively.
