Advanced Analytics with PySpark

Chapter 4. Making Predictions with Decision Trees and Decision Forests

Classification and regression are the oldest and most well-studied types of predictive analytics. Most algorithms you will likely encounter in analytics packages and libraries are classification or regression techniques, like support vector machines, logistic regression, neural networks, and deep learning. The common thread linking regression and classification is that both involve predicting one (or more) values given one (or more) other values. To do so, both require a body of inputs and outputs to learn from. They need to be fed both questions and known answers. For this reason, they are known as types of supervised learning.

PySpark MLlib offers implementations of a number of classification and regression algorithms. These include decision trees, naïve Bayes, logistic regression, and linear regression. The exciting thing about these algorithms is that they can help predict the future—or at least, predict the things we don’t yet know for sure, like the likelihood you will buy a car based on your online behavior, whether an email is spam given the words it contains, or which acres of land are likely to grow the most crops given their location and soil chemistry.

In this chapter, we will focus on a popular and flexible type of algorithm for both classification and regression (decision trees) and the algorithm’s extension (random decision forests). First, we will understand the basics of decision trees and ...