Unlike discriminant analyis that seeks to classify observations into known groups, cluster analysis seeks to identify hidden or suspected groups to which observations are thought to belong. Cluster analysis is a primarily inductive technique that infers the existence of clusters and employs various mathematical and statistical techniques to identify their nature.

Important Uses - With regard to Hong Kong's English language industry cluster analysis can prove useful in several ways

• Market segmentation - It can reveal who among the general public uses or does not use the English language
• Market behavior and attitudes - It can help to identify attitudes about the language held by different groups.
• Identifcation of test markets - It can identify different groups of users and thus provide greater focus for further investigation.

Cluster Types

• Well-defined Clusters - Clusters which exhibit both strong external isolation and good internal cohesion are well-defined.
  
  
  • External isolation - Members of the same cluster differ from those of another cluster by a fairly well-defined empty space. The criterion of external isolation is not well-suited for fuzzy clusters.
  • Internal cohesion - Members of the same cluster are similar and not separated by large empty spaces.

Classification techniques (analytical routines) more...