Planners and operators of shopping centers have to deal with two main challenges: The tenant mix (shops of specific retail categories) and the arrangement of tenants (category concentration / clustering or dispersion). The question if shops of the same retail category should be placed spatially concentrated or dispersed within a shopping center, has not been answered satisfactorily yet and actually there is reason to assume that it cannot be answered in general. A first step in empirical research on this question is the quantitative assessment of category concentration. Common methods of measuring the degree of clustering result in global measures without identifying single clusters at all or they lack theoretical foundation. This paper suggests the use of variable clumping method in order to characterize the empirical situation of clustering within a shopping center. It enables the identification of statistically significant clusters on different spatial scales as well as the pinpointing of the actual shops constituting these clusters. The procedure and its results may be of importance for both, practitioners and theorists. Academic researchers have tried to identify factors, which influence the success of a shopping center. The degree of concentration or dispersion is one of these factors and can be analyzed in a more profound manner by means of the variable clumping method. The application of the variable clumping method can enable further insights concerning research activities which dealt with the coherence of retail category concentration and structural features of shopping centers. Planners and operators can use the methodology in order to analyze the situation in their centers and to identify potential improvements according to scientific theories.

The paper gives some compendium of current research on retail concentration in shopping centers, explains the general principle of the variable clumping method and presents some exemplarily results for a medium sized German shopping center. Finally with the mean-k-nearestneighbor method, we introduce another new method which facilitates the analysis of concentration tendencies on a more abstract level. The paper aims at initiating further research including the effect of clustering on rents and customer flow distribution.