Main Content

Cluster Analysis and Anomaly Detection

Unsupervised learning techniques to find natural groupings, patterns, and anomalies in data

Cluster analysis, also called segmentation analysis or taxonomy analysis, partitions sample data into groups, or clusters. Clusters are formed such that objects in the same cluster are similar, and objects in different clusters are distinct. Statistics and Machine Learning Toolbox™ provides several clustering techniques and measures of similarity (also called distance metrics) to create the clusters. Additionally, cluster evaluation determines the optimal number of clusters for the data using different evaluation criteria. Cluster visualization options include dendrograms and silhouette plots.

Anomaly detection is a branch of machine learning that identifies observations that deviate from an expected pattern or distribution in sample data. Statistics and Machine Learning Toolbox provides several techniques for outlier and novelty detection (see Unsupervised Anomaly Detection), and additional methods for detecting anomalies in streaming data (see Incremental Anomaly Detection Overview).

Cluster Analysis Basics

Click to go to the example, Cluster Gaussian Mixture Data Using Hard Clustering