Fuzzy Geographically Weighted Clustering

Geodemographic analysis has been described as “the analysis of spatially referenced geodemographic and lifestyle data” (See and Openshaw, 2001, p.269) It is widely used in the public and private sectors for the planning and provision of products and services. Geodemographic analysis often uses clustering techniques which are used to classify the geodemographic data into groups, making the data more manageable for analysis purposes. Clustering identifies a number of geodemographic groups (clusters), each group having a particular geodemographic profile. Each geographical area under consideration is then assigned to a group based on its similarity to the group profile. Fuzzy clustering offers a method of clustering that uses the principles of fuzzy logic to calculate a membership value for each subject in each of the groups. So rather than assigning a geographical area to a single group, each area is allocated a membership value in each of the groups (clusters), thus helping to overcome the issues of ecological fallacy. The fuzzy clustering algorithm typically used in geodemographic analysis is Bezdek's fuzzy c-means clustering algorithm known as FCM (Bezdek et. al., 1984). Fuzzy geodemographic analysis using FCM has been investigated by Feng and Flowerdew (1998, 1999), and See (1999), but has received scant attention since an exception being the recent investigation by one of the authors (Mason, 2006). This paper proposes a modification to the fuzzy clustering algorithm to incorporate geographical effects, suitable for geodemographic analysis.