Machine Learning in Production: Developing and Optimizing Data Science Workflows and Applications

10. Classification and Clustering

10.1 Introduction

Classification algorithms solve the problem of sorting items into categories. Given a set, N, of samples composed of features, X, and a set, C, of categories, classification algorithms answer the question “What is the most likely category for each sample?”

Clustering algorithms take a set of objects and some notion of closeness and group the objects together using some criterion. Clustering algorithms answer the question “Given a collection of objects and relationships between the objects, what is the best way to arrange them into groups, or clusters, to satisfy a specific objective?” That objective could be compactness of the groups, in the case of k-means clustering. It might be making ...