Cluster analysis, or clustering, is the task of grouping a set of objects in such a way that objects in the same group (called a cluster) are more similar to each other than to those in other groups. It is a primary task of exploratory data analysis and a common technique in machine learning. Unlike supervised classification, clustering is an unsupervised task, meaning it does not use predefined labels. The goal is to achieve high intra-cluster similarity and low inter-cluster similarity. Various algorithms exist, based on different models of what constitutes a cluster, such as connectivity-based (hierarchical), centroid-based (k-means), or density-based (DBSCAN) models.
The formal study of cluster analysis originated in anthropology and psychology in the 1930s. It was first used by Driver and Kroeber in 1932 to study cultural traits and later introduced to psychology by Joseph Zubin and Robert Tryon. These early methods laid the groundwork for the statistical and computational techniques that are widely used today across many scientific and commercial fields.
Clustering is now a fundamental technique in data science and is applied in numerous fields. In marketing, it is used for market segmentation to identify distinct customer groups. In biology and bioinformatics, it helps classify genes with similar expression patterns and group homologous sequences. In computer science, it is essential for image segmentation, search result grouping, and anomaly detection. It is also used in social network analysis to identify communities and in finance for risk analysis.