Geography Reference
In-Depth Information
with national totals, and as noted above, clicking on points in these graphs takes users to
a map of the rate at the relevant date, centred on the current unit. The graphs are either
line or bar charts depending on the value of the 'continuous' attribute of the rate. Users are
currently able to change the comparison unit only with the modern units covered by the
2001 census, as essentially the same data are available for all of these; so for example you
can compare any pair of current local authorities.
However, most of the system's graphical repertoire is for presenting nCubes. Each admin-
istrative unit in the system, from the UK down to individual parishes, has its own 'home
page', which lists the statistical themes for which data are available. The theme pages be-
gin with links to the available rates, include some introductory text, and end with a list
of available nCubes, including the period they cover and the variables they contain. Three
alternative views are available for each nCube: a
Table View
lists the actual numbers; a
Source
Info View
provides information on sources, and for census data may include links to re-
constructions of the original table driven by the SDS; but the initial view of each nCube is
agraph.
Figure 13.6 shows how the system decides what kind of graph to create based on what it
knows about the relevant nCube. The very first case is slightly silly, as a one-dimensional
nCube based on a single category variable holds just one number: we present it as a one-
bar chart, always the same height but of course with a different scale. This graph, and all
other graphs we generate, is created by an existing graphing servlet rather than one written
especially for the project: we use Ce:wolf, based in turn on the Jfreechart class library.
Unlike the statistical mapping servlet, this does not extract data from Oracle for itself but is
passed an array of data by our software. Ce:wolf suffers from poor documentation but it is
very fast.
This approach works well for a variety of different kinds of nCube based on single variables,
whether for single dates or for time series, and also for two-dimensional nCubes where one of
the variables has only two categories. Most often with social statistics that variable is sex, but
Vision of Britain
always generates 'population pyramids' for nCubes of that structure. With
anything other than one-dimensional nCubes, time is always handled simply by creating
a sequence of separate graphs, one for each date. Presentation of more complex two- and
three-dimensional nCubes is less satisfactory, especially when data are available for a series
of different dates, and the system currently makes no attempt to graph nCubes with four
or more dimensions; if we had any, the current software would conceal their existence from
users. However, this is a limitation not of the overall approach but of a system with limited
interactivity where everything is done server-side.
13.4.3 Future potentials
The current graphical repertoire is severely restricted both by the rules imposed by the
funding body and our own limited resources to develop software. New funding from the
Joint Information Systems Committee is primarily aimed at extending our substantive
coverage to include British parliamentary election results from 1832 onwards, a large part
of the project being mapping the changing boundaries of constituencies. However, we will
be able to make some additions to our graphing and mapping capabilities so it is useful to
discuss the potentials.
Search WWH ::
Custom Search