Also available as an Acrobat File |

Editorial Assessment |
The Visualisation of Area-based Spatial Data ## 4.3 SAGE## 4.3.1 FunctionalityIn terms of functionality SAGE includes a balance of graphical and numerical methods and provides some tools for exploring the smooth and rough properties of both the spatial and non-spatial elements of the data. The usual range of numerical descriptive statistics (both resistant and non-resistant) can be output to a text window. Graphical tools for exploring the non-spatial elements of the data include boxplots, histograms and scatterplots. Spatial graphical tools include a Moran plot, in which the values for each area are plotted against the mean values for neighbouring areas, a type of trellis diagram based on spatial order and facilities for spatial filtering. SAGE is unique in providing facilities for classification (beyond that needed for cartographic display) and region building. Despite the known problems of area-based data and the sensitivity of results to the areal partition, this is one area of functionality which has received very little attention in the ESDA literature. SAGE also includes facilities for constructing and modifying a wide range of weights matrices, and making adjustments for variable base populations. There is no facility for handling missing values. SAGE has a spatial modelling capability, similar to SpaceStat, so that ESDA and other tools can be used for model assessment. All the graphical windows are linked, so that selecting an item in one automatically highlights the same item in all the other graphical windows. The statistical graphs are easy to read and contain axis labelling. Although it is possible to open multiple graph windows, tt is only possible to have one map window open, and this is drawn by the ARC/INFO GIS package. Altering the appearance of the map is done by using a series of ARC/INFO menus, and is rather long-winded and cumbersome. The default colour scheme is well suited to displaying categorical data, but not to data on rates or proportions. There is a limited spreadsheet capability that allows the user to see the non-spatial data, and to create new functions of the existing variables. ## 4.3.2 Ease of UseSAGE is controlled using a menu interface on the main SAGE window (top left window on the diagram). The menus can sometimes be non-intuitive, the worst example being in the drawing of graphs. After selecting the graph type from the main menu, the user is prompted to enter a name for the graph - it is not clear why this is needed, and it is not possible to allow the name to default - something must be typed even if it is nonsense. After this a blank window appears! In fact this is a canvas on which one or more graphs of the desired type can be drawn - in order to draw a graph the Add option on the window menu is selected, and this finally brings up a menu where the relevant variables can be selected. The documentation is such that it is fairly easy to learn how to use the system. However, given that some of the facilities provided are quite sophisticated, and really require the user to understand something about spatial analysis, and the use of weights matrices, it would take a beginner some time to become familiar with all the facilities. ## 4.3.3 ImplementationSAGE was unusual in this exercise, since it did not need to be downloaded or installed, as it was written at Sheffield. However, we know from the experience of those elsewhere who have tried to dowload the software that this presents problems. SAGE is linked to a particular version of the Unix operating system on Sun workstations, and does not appear to work on any other configuration. Since there is no longer any manpower available to provide technical support for it, this unfortunately limits its availability to other researchers. SAGE is written as a self-contained package, with no facility for additional facilities. It comes supplied with source code (in C++) but only a skilled programmer would be able to use this to modify the package significantly. SAGE reads its data from ARC/INFO, which means it does not have any data input facilities of its own. A complication, however, is that before a dataset can be read, the user must create a View file - this is in fact another ARC/INFO facility which SAGE simply makes use of, but is rather complex for a novice, and as the first thing which must be done, rather a daunting introduction to the package. |

Graphics Multimedia Virtual Environments Visualisation Contents