Search Filter
- Publications
- Abstract of Theses and Dissertations
- Database
- Some Applications of Cluster Analysis
Some Applications of Cluster Analysis
Thesis Abstract:
The study attempted to illustrate a method of hierarchical clustering, namely, furthest neighbor clusted analysis, and some of its applications in social and biological sciences; to explain some methods of standardizing variables as a way of equalizing each variable’s contribution prior to the clustering process; and to introduce some SAS (Statistical Analysis Systems) program packages, especially the CLUSTER procedure and the MATRIX procedure, with their capability in data manipulation.
The data set used to illustrate the application of cluster analysis in social sciences consisted of 118 data units (countries), each recorded by 16 variables (social indicators for year 1970). The data set used to illustrate the application of cluster analysis in biological sciences was composed of 65 data units (Vietnamese rice varieties), each recorded by 37 variables (varietal characteristics).
Depending on the nature of the variables, the data matrix of each data set was properly standardized. On the basis of these standardized variables, cluster analysis was used to group the data units of each data set into the so-called “natural groups” or “clusters.” For each data set, the hierarchical tree called “dendogram” was then drawn to display clustering results.
For both data sets, clustering results showed that cluster analysis could be used as a very useful tool in classification problems.