Linked Activity Spaces: Embedding Social Networks in Urban Space - Computational Approaches for Urban Environments - page 319

number of social triangles (green circles as activity space centroids) form a tight
linear cluster in the core area. This pattern is statistically significant, showing
high Getis-Ord GI* Z scores with p -values > .0001 in the area slightly east of the
downtown (in light color, Fig. 13.9 ). P -values are not significant in other parts
of the city. Interestingly, centroids covering this neighborhood also saw the most
significant modularity clusters (Fig. 13.8 ).
We also hypothesized that peripheral areas may be more mixed. We reject this
hypothesis as the peripheral areas seem to host more cohesive communities than the
downtown area. The POIs shared by linked activity spaces are more frequently on
the periphery of the city. A POI can have from .01 % to 2.8 % of its pairs shown to be
friends. A POI with a high value indicates that it is located in a convenient area for
linked activity spaces (i.e. friends) to meet, where POIs with a lower value is more
likely to host non-friend individuals. The POI hosting 2.8 % linked activity spaces
is located near many popular hotels and the city's largest park - a notable tourist
site in the city. This site might be a popular meeting spot for friends, but it may also
be the artifact of many local business calls to one another in nearby buildings.
More generally, this ratio increases further from the city center (Figs. 13.10 and
13.11 ), so that POIs on the outskirts of the city are more likely to host pairs of
contacts. This may be because many people have activity spaces that stretch into
the downtown for work, but have their social contacts closer to their residences
in another part of the city. Though agents with more contacts tend to frequent the
dense downtown, given a POI with 100 random people, a user is more likely to find
a friend within this 100 people if he or she is at a suburban POI, than an urban

POI 'Familarity' and Distance from City Center

100 500 1000 5000 10000

Distance of POI to City Center (Log−scale Meters)

Fig. 13.10 As the POIs are located farther away from the city center there is a higher ratio of
friend to non-friend pairs using these points