Attribute space visualization of demographic change

Abstract

This paper introduces an approach for closer integration of self-organizing maps into the visualization of spatio-temporal phenomena in GIS. It is proposed to provide a more explicit representation of changes occurring inside socio-economic units by representing their attribute space trajectories as line features traversing a two-dimensional display space. A self-organizing map consisting of several thousand neurons is first used to create a high-resolution representation of attribute space in two dimensions. Then, multi-year observations are mapped onto the neural network and linked to form trajectories. This method is implemented for a data set containing 254 counties and 34 demographic variables. Various visual results are presented and discussed in the paper, from the visualizations of individual component planes to the mapping of voting behavior onto temporal trajectories.