Decision trees are the first nonlinear model in the course and the building block for the two most powerful classical methods in finance: random forests and gradient boosting. A tree asks a sequence of yes-or-no questions about the features and follows the answers down to a prediction. It is intuitive, handles nonlinear interactions for free, and needs almost no data preprocessing. This lesson explains how a tree is grown, the splitting criteria that decide each question (Gini, entropy, and variance reduction), works a split selection by hand, and shows why a single tree overfits and how to control it. We close with the strengths and weaknesses interviewers probe and the path to ensembles.
Table of Contents
Start Your Interview Preparations!
Already have an account? Log in!