Summary

In this chapter, we discussed hierarchical clustering and DBSCAN, and in what type of situations they are best employed. While hierarchical clustering can, in some respects, be seen as an extension of the nearest-neighbor approach seen in k-means, DBSCAN approaches the problem of finding neighbors by applying a notion of density. This can prove extremely beneficial when it comes to highly complex data that is intertwined in a complex fashion. While DBSCAN is very powerful, it is not infallible and can even be overkill, depending on what your original data looks like.

Combined with k-means and hierarchical clustering, however, DBSCAN completes a strong toolbox when it comes to the unsupervised learning task of clustering your data. When faced with any problem in this space, it is worthwhile comparing the performance of each method and seeing which performs best.

With clustering explored, we will now move onto another key piece of rounding out your skills in unsupervised...