Methodologies of officer billet classification.

Abstract

Four potentially valuable methods to classify officer billets into subgroups on the basis of multivariate observations about the billets are presented. The methods aiming to reduce the dimensionality and to identify homogeneous subgroups are: Principal Component Analysis, Multidimensional Scaling, Hierarchical Cluster nalysis (Hiclust) and Cluster Analysis Optimizing an Objective Function (K-Means) . They are applied to a data set obtained from an outside source and comprising 96 Navy officer billets. Thirteen quantitative variables measuring the relative amount of time spent for managerial responsibilities and resources have been entered into the analysis. On the basis of the entered variables, the presence of eight billet clusters have been determined. The evolved groups are described by their centroids and within-group standard deviations.