Geoscience Reference
In-Depth Information
In working toward this goal, those looking to examine social/spatial problems are
aided by the recent proliferation of large datasets evidencing human social contact
and movement (such as GPS or cell tower usage records) in the city (Reades et al.
2007
). The integration of human movement and activity data, such as information
from GPS traces (Gao et al.
2013
), check-in data (Cho et al.
2011
), online social
networks (Scellato et al.
2011
), and photo-sharing sites (Crandall et al.
2010
;
Girardin et al.
2008
; Sun et al.
2013
), into urban models are providing new windows
on how humans use the built environment. Specifically, the use of mobile telephone
calls to understand city usage patterns are becoming a cornerstone of modern urban
informatics, planning, and transportation (Ratti et al.
2006
). We take advantage
of mobile telephone call data to test our research questions about the locality or
dispersion of social ties in the city.
Further, the relative convenience of colocation for friends can be evaluated.
Calabrese et al. (
2011
) find that in 94 % of telephone calling partners, one partner
constantly travels further to meet. On average, the partner traveling further travels
3 times further to meet. This method uses travel time and distance, which is
important for logistics. However, we extend this concept by incorporating the built
environment into these compromises, to show where in the city friends are likely
to meet. By spatially-linking the respective activity spaces of two friends in the
GIS, we can better understand how the city is able to provide places for friends to
meet, and assess the travel needs to do so—i.e. it is relatively easy for friends with
spatially-overlapping activity spaces to meet face-to-face.
13.1.2
Linked Activity Spaces
We use cell phone call data records (CDRs) to model “friendships” (i.e., interper-
sonal relationships) as a social network, inferred by the frequency calls between
two agents, and the sets of locations visited by each member of the social network
within the city (i.e., activity spaces). A pair of activity spaces of an ego and alter
are called
linked activity spaces
(LAS) if the ego and alter are friends (i.e., contacts)
in the dataset. The two activity spaces of friends are modeled within the GIS and
spatially analyzed for similarity, via the number of “third places” shared among the
pair (following Rosenbaum
2006
). Moreover, we analyze the social network as a
whole to find whether high-degree
egos
(a.k.a. those with many friends),
triangles
(groups of three agents) and
communities
use the city in significantly similar ways.
We have four main hypotheses for the analysis of LAS. (1) We expect that
friends' activity spaces will overlap more often than a random pair of activity spaces,
indicating that friends use the city more similarly than a random pair of people.
(2) We also hypothesize that egos with high degrees or high clustering coefficients
(see Jackson
2010
) will be more associated with the city center, as this denser
environment tends to have more meeting places, diverse services, commercial areas,
and nightlife. (3) In terms of city form and groups, we believe that central areas will
Search WWH ::
Custom Search