|Also available as an Acrobat File|
The Visualisation of Area-based Spatial Data
Like cdv, MANET has been designed principally to provide facilities for the graphical exploration of data, with the additional emphasis on the development of methods for dealing imaginatively with missing values (hence the name of the package - Missings Are Now Equally Treated).
In terms of our classification of ESDA functionality, MANET has excellent facilities for exploring the non-spatial elements of data. These include standard graphs, such as histograms, boxplots and scatter plots, and more innovative techniques such as trellis plots, in which a series of separate boxplots are drawn for different categories of data (e.g. data from separate years) in a tabular form (an example is shown in the central window of the screen layout above). The trellis plot is actually an example of what Tufte (1990) calls `small multiples', in which repeated graphics are used to encourage comparison between different subsets of data, and would make it extremely easy to detect differences between different categories within a dataset. There are some very useful additional features including weighted and `spine' versions of histograms, and the capability to automatically and continuously change the histogram bin sizes.
Exploration of the spatial elements of the data is done using the linkage between the map and the graphs. For example, by moving a selection tool across a graph, the geography of the data set can be explored. However, there appear to be no specialist spatial graphs or tools for detecting spatial properties such as trends or clusters.
All the windows are linked to one another, such that selecting an element in one (e.g. a point on a scatter plot) highlights the related elements in the others (e.g. the area on the map). The selection of elements can be done in numerous ways, selected using a special graphical menu - the default is for selection of a single element at a time. However, it is possible to set the system up so that selections are cumulative for example.
A characteristic feature of the package is that most of the displays do not show the actual values of individual data items - indeed it is very difficult to discover what the data values are at all, since there is no tabular display of the data matrix, and no tools for producing summary statistics. The focus of the software is thus entirely on the detection of patterns and trends.
4.2.2 Ease of Use
MANET is extremely easy to use, once the basic operation of the system is understood. Some of the newer features (such as drawing a map of the data) are not yet documented, so that assistance was required from the author to test these out. The manual is clear, apart from the description of the various options for selecting cases. This caused a good deal of confusion until an explanation was discovered on the Web page.
MANET is available by email from the author and was downloaded and installed without difficulty.
Data can be input using a simple text file format.
Graphics Multimedia Virtual Environments Visualisation Contents