Summary

In this chapter, you learned a lot about the unsupervised learning technique called cluster analysis. It helps you when you do not have a response variable but you believe that there are natural groupings in the data. There are many types of clustering algorithms, and you learned two. K-means clustering is widely used and is ideal when you have constraints or a sense of how many clusters exist in your data. It is straightforward to implement and you can pull elements out of the model to perform other analysis. You used k-means to determine the best number and location of customer service kiosks. Hierarchical clustering is a good choice when you do not have a sense of the number of groups that may exist in the data. You used this to perform customer segmentation of two-dimensional demographic data. You learned how to use the elements from k-means to help evaluate the right number of clusters to select, as well as visualize the output of hierarchical clustering.

In the next chapter...