In this chapter we presented some of the most important unsupervised learning methods. We did not intend to provide you with an exhaustive introduction to all the possible methods, but instead a brief introduction to these kinds of techniques. We described how we can use unsupervised algorithms to perform a quick data analysis to understand the behavior of the dataset and also perform dimensionality reduction. Both applications are very useful as a step before applying a supervised learning method. We also applied unsupervised learning techniques such as k-means to resolve problems without using a target class—a very useful way to create applications on top of untagged data.

In Chapter 4, Advanced Features, we will look at techniques that will allow us to obtain better results in the application of machine learning algorithms. We will look at data-preprocessing and feature-selection techniques to obtain better features to learn from. Also, we will use grid search techniques to obtain the parameters that produce the best performance with our algorithms.