Introduction

In previous chapters, we evaluated a number of different approaches to data clustering, including k-means and hierarchical clustering. While k-means is the simplest form of clustering, it is still extremely powerful in the right scenarios. In situations where k-means can't capture the complexity of the dataset, hierarchical clustering proves to be a strong alternative.

One of the key challenges in unsupervised learning is that you will be presented with a collection of feature data but no complementary labels telling you what a target state will be. While you may not get a discrete view of what the target labels are, you can get some semblance of structure out of the data by clustering similar groups together and seeing what is similar within groups. The first approach we covered to achieve this goal of clustering similar data points is k-means. K-means clustering works best for simple data challenges where speed is paramount. Simply looking at the closest data...