A COMPARISON OF FACTOR AND CLUSTER ANALYSIS

Abstract

Statistical methods are frequently used both in scientific studies and in daily life. However, it is very important to choose the correct statistical method and to make interpretations accordingly to this. Among the multivariate statistical methods factor analysis and cluster analysis seem to be similar methods in terms of grouping.

Factor analysis is generally used when wondering whether a large number of variables can be analyzed with a smaller number of basic variables. It is a method that decides whether to gather a large number of variables under several headings and renames data collected under certain headings. Thus, as the dimension is reduced and the number of variables is reduced by factor analysis, these variables are also classified. 

When the word cluster is mentioned, the community consisting of similar objects comes to mind. Cluster analysis is a multivariate statistical analysis technique used to divide similar observations and sometimes similar variables into homogeneous subgroups. Although cluster analysis is generally used to cluster observations, it is stated that it can also be used to cluster variables. However, in splitting variables into clusters, factor analysis is mostly used. Extreme values can also be determined with cluster analysis. As a result of the analysis, an observation that results as if it formed a cluster alone that could not be included in any cluster can be considered an extreme value. Cluster analysis used in grouping the observations may also be likened to discriminant analysis. However, in cluster analysis, the groups to be obtained as a result of the analysis are made at the beginning of the study in discriminant analysis. 

In this study, the factor analysis, which is mostly used to group variables, and the cluster analysis methods used to group the observations were compared.