Clustering

We use clustering to divide our data points into groups of similar points. The points in each group are more like their fellow group members than those of other groups. Clustering is commonly used for tasks such as recommendation systems (think of how Netflix recommends what to watch based on what other people who've watched similar things are watching) and market segmentation.

For example, say we work at an online retailer and want to segment our website users for more targeted marketing efforts; we can gather data on time spent on the site, page visits, products viewed, products purchased, and much more. Then, we can have an unsupervised clustering algorithm find groups of users with similar behavior; if we make three groups, we can come up with labels for each group according to its behavior:

Figure 9.17 – Clustering website users into three groups

Since we can use clustering for unsupervised learning, we will need to interpret...