Linked Activity Spaces: Embedding Social Networks in Urban Space - Computational Approaches for Urban Environments - page 321

First, the LAS method allows us to find the extent to which a pair of friends is
more likely to use the same places in the city more than a random pair, so that a
user i is more likely to have frequented the same POIs as a friend j than a random
user r .
Second, we find that egos with high degrees are inclined toward the city center,
while egos with high clustering coefficients show no significant spatial correlation.
These results are not necessarily intuitive, as community members in suburban areas
might also have high degrees, but do not seem to. Also, users who cling to the
urban core (or tight suburban neighborhoods) might also be expected to be part of
a number of “cliques” or friend groups; however, this also does not seem to be the
case.
Third, we expected that central areas would play an enhanced role in supporting
“clique-like” and modular groups, but we found, counter to our expectation, that
the downtown was indeed a mixing pot for many groups. We do, however, find two
specific neighborhoods that tend to harbor enclosed (“clique-like”) social groups.
Moreover, triads of friends are likely to use the downtown area.
Also, we had expected that peripheral areas were more mixed as suburbanites
often have more access to automobiles and, thus, may not choose to live next to
their contacts if they can drive to other parts of the city to visit. Counter to our
initial hypothesis, peripheral areas show less frequent mixing of social groups and
friends than any other part of the city.

13.5.1 Utility

We find these results useful for theoretical and practical issues in planning. First, in
Jiamusi, we find that the ratio of friends' shared POIs to random pairs' shared POIs
is 59:46. This ratio can be considered an indicator of clustered socialization. When
high, this ratio shows that friends tend to group in certain parts of the city, or use the
same amenities in a city. This ratio represents an urban feature that can be compared
across cities, over time, and as it correlates with urban features such as population,
crime rates or traffic.
Next, a planner can use these findings, for example, to discuss the merits
of different models of urbanism. For example, while New Urbanism focuses on
neighborhood design, architectural style, and transit-oriented development for high-
density walkable cities (Al-hindi and Till 2001 ; Vanderbeek and Irazabal 2007 ), and
Landscape Urbanism argues that urban design should be flexible and open-ended,
(Waldheim 2002 ), by leveraging existing resources (Cranz and Boland 2004 )and
preserving wilderness (Yu et al. 2011 ). The LAS method results can be used to
probe the adverse consequences of urban sprawl, such as its challenges for social
life (Gehl 1987 ).
This analysis can be used to plan the location of third places (Rosenbaum 2006 ),
such as restaurants, parks, coffee shops, theaters, and other facilities. Locations