Analyzing Clustering Effects across Time
Douglas M. Dunn, James M. Landwehr · Journal of the American Statistical Association · 1980
Several graphical statistical methods tailored particularly for clustering applications are described and discussed. Certain techniques give ways of examining the internal cohesiveness of clusters, both across all variables and in terms of particular variables. Other graphical methods are concerned with relationships within and between clusters and with variables exogenous to the clustering. The methods focus on changes in cluster characteristics across two successive time periods. These techniques are illustrated with data from a cluster analysis using economic characteristics of 48 firms from one industry.