Tree-Based Methods

Abstract

This chapter explores tree-based methods for demand prediction. These methods are widely used given their strong predictive power. We consider three types of methods: Decision Tree, Random Forest, and Gradient Boosted Tree. We apply these methods under both the centralized and decentralized approaches. For each method, we briefly discuss the underlying mathematical framework, present a common practical way to select the parameters, and detail the implementation process by providing the appropriate codes. We conclude by comparing the different methods in terms of both prediction accuracy and running time.