Random Forest

This one caught me off guard, 'random forests' what the heck? What does something related to data would be called forest. Then I learned that random forest is nothing but a collection of decision trees, get it? Many trees together... forest, that's some clever name.

A decision tree is when make a decision based on yes/no questions which are answered at each level of the decision tree (example below). Decision trees can be ensembled together so that a goal can be ran through many decision trees in order to improve model score and more accurate predictions.

Example:

Goal: Predict if a person will buy a car.

Questions (splits):
1. Is income > $100K?
 → Yes → next question
 → No → predict “No”
2. Is age < 30?
 → Yes → predict “Yes”
 → No → predict “No”