The 21st century is the digital century, where every person on every rung of the economic ladder is using digital devices and producing data in digital format at an unprecedented rate. 90% of data generated in the last 10 years was generated in the last 2 years. This is an exponential rate of growth, where the amount of data is increasing by 10 times every 2 years. This trend is expected to continue for the foreseeable future:

Figure 1.1: The increase in digital data year on year

But this data is not just stored in hard drives; it's being used to make lives better. For example, Google uses the data it has to serve you better results, and Netflix uses the data it has to serve you better movie recommendations. In fact, their decision to make their hit show House of Cards was based on analytics. IBM is using the medical data it has to create an artificially intelligent doctor and to detect cancerous tumors from x-ray images.

To process this amount of data with computers and come up with relevant results, a particular class of algorithms is used. These algorithms are collectively known as machine learning algorithms. Machine learning is divided into two parts, depending on the type of data that is being used: one is called supervised learning and the other is called unsupervised learning.

Supervised learning is done when we get labeled data. For example, say we get 1,000 images of x-rays from a hospital that are labeled as normal or fractured. We can use this data to train a machine learning model to predict whether an x-ray image shows a fractured bone or not.

Unsupervised learning is when we just have raw data and are expected to come up with insights without any labels. We have the ability to understand the data and recognize patterns in it without explicitly being told what patterns to identify. By the end of this book, you're going to be aware of all of the major types of unsupervised learning algorithms. In this book, we're going to be using the R programming language for demonstration, but the algorithms are the same for all languages.

In this chapter, we're going to study the most basic type of unsupervised learning, clustering. At first, we're going to study what clustering is, its types, and how to create clusters with any type of dataset. Then we're going to study how each type of clustering works, looking at their advantages and disadvantages. At the end, we're going to learn when to use which type of clustering.