R Data Analysis Cookbook, Second Edition - Second Edition

Performing Advance clustering

Both hierarchical clustering and partitioning methods (K-means and PAM clustering) are suitable to find spherical-shaped clusters or convex clusters and they use a heuristic approach to construct clusters, but they do not rely on a formal model and are severely affected by the presence of noise and outliers in the data. In this recipe, we will look at two advance clustering techniques to handle these shortcomings--density-based clustering and model-based clustering.