|Also available as an Acrobat File|
The Visualisation of Area-based Spatial Data
Stephen Wise, Robert Haining & Paola Signoretta
Sheffield Centre for Geographic Information and Spatial
Area-based data, such as from the decennial census of population, are an extremely important source of information for many social science disciplines. Such data have a number of characteristics which make them different from other data used in the social sciences, such as the sensitivity of analytical results to the size and arrangement of the area boundaries, and this has lead to the development of specialist software tools for their analysis and display.
This case study presents a review of the current state of the art of software for the visualization of area data. The focus is on the use of visualization methods for exploratory spatial data analysis (ESDA) which normally encompasses one or more of the following: (1) the identification of unusual data values or errors in the data, (2) the detection of patterns in the arrangement of data values, (3) hypothesis formulation from the data and (4) the assessment of models of the data.
Four packages were selected for review based on their availability within the timescale of the work, and the range of visualization tools they offer. Two elements were found in common in all four packages:
These facilities provide an extremely flexible environment for visualising the non-spatial elements of the data - both the `smooth'characteristics such as the distributions of variables, and the `rough' characteristics such as outliers from the distribution.
However there was a difference in the extent to which the packages provided tools to identify the explicitly spatial characteristics of the data. In two cases, the emphasis was on the provision of flexible graphical facilities, allowing extremely quick and easy redrawing of mapped views of the data, and graphical exploration of the location on the map of cases identified from the other graphs. The other two packages provided a series of facilities specifically for spatial data, such as the drawing of boxplots for areas at defined spatial lags, spatial smoothing to identify trend and the calculation of measures of local spatial autocorrelation and spatial concentration, for the detection of clusters of high or low data values.
We consider that both approaches are valuable, and that the development of a software package which combined the strengths of both, and was easily available to the academic community would be a useful tool for those wishing to analyze area-based spatial data. . In the report, the functionality which should a package should contain is identified, drawing on the examples of best practice identified in the four packages reviewed. Such a package would need to be easy to implement on commonly used platforms, and come supplied with documentation, including an introductory guide which outlined the main facilities, and explained the user interface. Given the importance of ease of implementation and documentation, it might be necessary to seek collaboration with a professional software vendor in the production of the software.
Despite the fact that graphical facilities are seen as playing such a key role in ESDA, little work has been done to assess the graphical effectiveness of visualization tools, and we identify this as one area for further work. In the report we outline one possible methodology, drawing on work from the field of statistical graphics, which could be explored further.
Work on visualization tools for area-based spatial data has taken place within the framework of existing statistical and cartographic software, rather than using existing ViSC software tools. It would be extremely interesting to see whether ViSC products could in fact be used for the visualization of area-based data, since such products are already available to the academic community and structures have been put in place to support them, and this is also suggested as a topic for future research.
Graphics Multimedia Virtual Environments Visualisation Contents