Hands-On Data Structures and Algorithms with JavaScript

The final information gain per branch

Now, we can calculate the entropy of all the branches using a frequency table of two attributes for the rest of the columns, similar to what we did for Outlook, and get the following result:

For the branch Humidity, there are two possible sub-branches, which have the following results breakup:

yes	no	total
Humidity	high	3	3	6
normal	6	1	7
13

 

Similarly, for Wind, the breakup is as follows: 

yes	no	total
Wind	weak	6	2	8
strong	3	2	5
13

For Temperature, it is as follows:

yes	no	total
Temperature	Hot	2	2	4
Mild	4	1	5
Cool	3	1	4
13

 

We calculate the branchEntropy and the Information gain for each branch, and here are the results following similar ...