It may be useful to compare the grade of fuzzy proximity membership function
used in the computation of the similarity score with other possible measures. Figure
4.13 illustrates the setting where the arrow that stretches outwards from the
geographic center of the locality of Buffalo is depicted as an x-axis in Figure 4.14.
The first option is to use distance as a measure of dissimilarity. Another option is
to consider buffers delineated around the Core.
To measure the distance between a polygon (locality) and a point (accident), a
representative point of polygon, such as the centroid, is required. This leads to a
loss of geographic detail. Moreover, distance option ignores the anomalous proximity
surface that is created by neighborhood relations. For instance, if the arrow
(in Figure 4.13) is drawn along a different direction, the membership values will
be adjusted to new neighborhood relations in the case of fuzzy proximity whereas
the distance option maintains the membership value regardless of the direction.
Creating multiple buffers around the Core does not require a centroid, but proximity
decreases in an abrupt manner along the buffer boundary. As for the distance