Often, a set of data can be organized into a set of clusters. For example, you may be able to organize data into clusters that correspond to certain underlying properties (such as demographic properties including age, sex, geography, employment status, and so on) or certain underlying processes (such as browsing, shopping, bot interactions, and other such behaviors on a website). The machine learning techniques to detect and label these clusters are referred to as clustering techniques, naturally.
Up to this point, the machine learning algorithms that we have explored have been supervised. That is, we have a set of features or attributes paired with a corresponding label or number that we are trying to predict. We use this labeled data to fit our model to the behavior that we already knew about prior to training the model.
Most clustering techniques are unsupervised...