As we have learned in the last chapter, data always contains valuable insights. Exploring with statistics, filters, and charts is a great tool for this. However, data has another internal value—its predictive power; it can be used to fit an algorithm (machine learning model) that will then be able to predict the values of interest and explain its judgment.

Machine learning (ML) is a large and complex topic that is clearly out of the scope of this book. Indeed, building an advanced and complex model requires deep theoretical knowledge of the specific domain and a lot of time and exploration. However, some ML models are very simple and easy to comprehend, and the basic underlying principles are all the same. Many ML models don't provide a good interpretation—but here, we'll use the ones that do.

In this chapter, we'll discuss the basics of ML and train a few basic ML models on our WWII dataset. We'll further interpret their behavior and caveats and how to mitigate some of the issues. In particular, we'll cover the following:

• Basics of ML
• Unsupervised learning (clustering) using the k-means algorithm
• Supervised learning with k-nearest neighbors and linear models