Hierarchical clustering

k-means clustering uses a top-down approach because we start the algorithm from the most important data points, which are the cluster centers. There is an alternative approach of clustering where, instead of starting from the top, we start the algorithm from the bottom. The bottom in this context is each of the individual data points in the problem space. The solution is to keep on grouping similar data points together as it progresses up toward the cluster centers. This alternative bottom-up approach is used by hierarchical clustering algorithms, and is discussed in this section.