
3.5 Cluster Analysis
■
47
3.5 Cluster Analysis
Cluster analysis attempts to isolate regions of similarity within a database
and find the relationships between multiple clusters. Large commercial
and governmental databases contain huge quantities of data. Data mining
explorers discover that such large repositories are not lacking in features,
patterns, and relationships but are instead saturated with patterns. There
are just too many. Directed data mining techniques, pursuing a specific
objective function, often find nothing but high levels of noise. It is here
that cluster analysis, in the form of automatic cluster detection (ACD),
provides a truly unsupervised ...