Examination of Several Results of Different Cluster Analyses with a Separate View to Balancing the Economic and Ecological Performance Potential of Towns and Cities

Abstract

The objective of this paper is to compare cluster analyses conducted by using different methods for 116 administratively autonomous municipalities (kreisfreie Staedte) in Germany. The cluster analyses aim to provide answers to the question as to the impact of land-use structures on the performance potential of towns and cities. Drawing on the database established, 11 attribute variables for the analysis were selected that significantly characterise a city’s land-use structures and go a long way towards moulding its economic and ecological performance. We show that no cluster structure exists in the data set. Therefore we investigate the data set by using Gaussian Mixture-Models estimating by Expectation-Maximization (EM) algorithm. This indicates that three or two variables suffice to classify the cities. The next step in our exploratory research is to conduct and to compare the results of different classification algorithms for these three and two variables. The classification based on EM algorithm allows us to identify 8 classes. We discuss them and compare this result with results of some cluster analyses with a separate view to balancing the economic and ecological performance potential of towns and cities.