Exam 2 Study Guide

Exam 2 will consist of approximately 3 multiple choice questions on each topic listed below:

• Comparison of different regression learner performance characteristics: Trees, forests, KNN, linreg
• Comparison of learner types: Regression, Classification, RL
• Overfitting: Definition, how to identify, what might prevent it, what might cause it?
• Bootstrap aggregating.
• Boosting.
• Decision trees. Random versus information based construction. Advantages of one over the other.
• Reinforcement learning: How is it defined? Questions about State, Action, Transitions, Reward
• Q-Learning. The update equation, definition of Q
• Dyna-Q
• Things you should know because you did the projects. In sample versus out of sample. Istanbul problem, why did shuffling help?
• Options

Readings:

• "Machine Learning", Chapter 1, Introduction
• "Machine Learning", Chapter 8, Instance-based Learning
• "Machine Learning", Chapter 3, Decision Tree Learning
• "Machine Learning", Chapter 3, Decision Tree Learning
• Paper: "Perfect Random Tree Ensembles" by Adele Cutler
• "Machine Learning", Chapter 13, Reinforcement Learning

The exam will include 30 questions. You will have 30 minutes (plus 5 extra in case of a problem) to complete the exam. The exam is closed book, closed notes. You can use an on screen calculator.