Summary
We sometimes need to organize our instances into groups with similar characteristics. This can be useful even when there is no target to predict. We can use the clusters created for visualizations, as we did in this chapter. Since the clusters are easy to interpret, we can use them to hypothesize why some features move together. We can also use the clustering results in subsequent analysis.
This chapter explored two popular clustering techniques, k-means and DBSCAN. Both techniques are intuitive, efficient, and handle clustering reliably.
