The geographical analysis of data presents a number of special problems. To tackle these problems, the project has made use of a Geographical Information System called MapInfo.
The tools within MapInfo enable users to create a number of different types of map. Of particular relevance to this project has been the creation of contour maps using the MapInfo “add-in” Vertical Mapper.
There are several reasons for generating there types of maps:
Data Protection
The Data Protection Act requires that living individuals cannot be identified. By using postcodes as a means of geographic referencing and then generating contours from the postcoded data the results are sufficiently anonymous to satisfying the Data Protection Principles.
A unit postcode (e.g. LE1 6ZG) in a suburban ward such as Evington will average 18 addresses compared with 26 in a council housing area like Eyres Monsell.
Geographical Independence
When data is collected, it is usually done so on the basis that the subjects are exclusively located within an administrative area of some size and shape. By dividing places into areas, arbitrary groupings are created which are known to influence how the aggregate characteristics of those areas appear. The use of postcodes provides a means of using very small clusters of people that they are assumed to be relatively homogenous while also being manageable in number. By this means it is assumed that the inherent patterns within the data will become more evident rather than being obscured by the effect of the artificial boundaries.
Clarity
Contour maps can simplify the data more effectively than plotting other types of geographies such as thematic maps of points. The latter example can have problems with the points overlapping each other thus obsuring values. Moreover, users are familiar with the concepts via such examples as TV weather maps.
The technique used to create the contours is known as “Inverse Distance Weighting”. This method creates a surface or grid of regularly spaced points. Each point has a value that is based on the average of the nearest surrounding values of the original data. This has the effect of creating a spatial moving average and so smooths out the more extreme values within the data to reveal the underlying patterns.
This approach is best suited to data which is highly variable such as information about the characteristics and behaviour of people.
The search and display parameters can be set independently and these can be used to control the sensitivity of the values of contours as well as control the extent to which the contours cover an area.
In these maps the search radius was set to 1000m which is about a sixth of the radius of the City. However, the display radius was set at 250m so that the parts of the City where people did not live were kept blank to differentiate them from residential areas.
