PATTERN RECOGNITION IN LARGE GEOGRAPHICAL DATABASES: TOWARDS A DETAILED ASSESSMENT OF THE WORLD CITY NETWORK

Abstract

Although a detailed empirical analysis of the world city network is essential to attain insight in its functioning, it can be noted that previous explorations have been restricted to analyses of a limited number of thoroughly connected cities. A major reason for the neglect of less connected nodes in this global urban network is the sparse evidence on their world city formation. Drawing on earlier specifications and measurements of the world city network, the present paper shows how fuzzy set approach and pattern recognition can assess the inherent vagueness in classifications of lower ranked world cities. The resulting taxonomy asserts the intertwining relational tendencies of 234 cities in 20 clusters. Key findings include the distinctive profiles of US cities, the marginal position of (sub-Saharan) African and Central American cities, and Miami's particular role as a gateway between Anglo- and Latin America.