Unsupervised learning tries to discover basic patterns in the data. It focuses on the data themselves, whereas supervised learning (see our following chapters) focuses on the underlying data-generating process. An example of unsupervised learning is exploratory data analysis (see our previous chapter). However, exploratory data analysis is often limited, especially when the samples (such as photons) are associated with more than just one or two features (not only wavelength and color, but also spin, direction, location, ...). We then need more elaborate methods. This chapter describes two such methods: principal-component analysis and k-means clustering.