Introduction to UL
We will define UL as a subset of ML in which models are trained without the existence of categories or labels. Unlike its supervised counterpart, UL relies on the development of models to capture patterns in the form of features to extract insights from the data. Let's now take a closer look at the two main categories of UL.
There exist many different methods and techniques that fall within the scope of UL. We can group these methods into two main categories: those with discrete data (clustering) and those with continuous data (DR). We can see a graphical representation of this here:
Figure 6.1 – The two types of UL
In each of these techniques, data is either grouped or transformed in order to determine labels or extract insights and representations without knowing the labels or categories of the data ahead of time. Take, for example, the breast cancer dataset we worked with in Chapter 5, Understanding Machine Learning, in...