ALGORITHM 1: Information gain pseudocode.

Input: Dataset D= { (x1,c1),…,(xN,cN) }

Output: predicted class Ci.

for each attribute or node i do

Calculate the information of each branch

Average information of all attribute values or branches, avg

Calculate the information of the unsplit node, unsplitINFO

Calculate the information gain of the node i:

Gain[node] = unsplitINFO — avg

end

return Gain[node]