Social area analysis, data mining, and GIS

There is a long tradition of describing cities through a focus on the characteristics of their residents. A brief review of the history of this approach to describing cities highlights some persistent challenges. To fit within the constraints of widely used multivariate data reduction techniques and thematic cartography populations are classified. Historically, this classification has been guided by theory (i.e. Shevky-Bell) or simply the desire to efficiently describe urban populations. The labeling of classes reduces the very complexity these multivariate methods and maps are trying to capture. This problem is addressed through a geodemographic approach that uses the Kohonen Self-Organizing Map (SOM) algorithm. A dataset describing 79 attributes of the 2217 census tracts in New York City is analyzed through a method that pairs a SOM with a GIS. The resulting “maps” represent social space and geographic space and can be used to explore the similarities and differences among regions of the city and the geographic patterns of different geodemographic classes. Pairs of social and geographic maps are formally compared using simple pattern metrics.