The methods used in the previous section created a surface
whereby the grid cells represent the hotspots based on the density
measure. The number of grid cells in a hotspot varies, showing the
hotspots are not uniform in size or shape. To be part of a hotspot
a grid cell has to have an accident density level which is over a
specified threshold, indicating these are the areas in London where
the accident density prevalence is at its most intense. This grid
surface provides the basis for collating the accidents which occur
within these grid basedhotspots. The result is a database of hotspots
whereby the nature of the database means that there is no analysis
of individual accidents, but an analysis of groups of accidents which
share a common nearby spatial location, implying a common casual
factor. By ascertaining the nature of this similarity, comparisons
between hotspots can be made on a ‘like by like’ basis.