Linked Activity Spaces: Embedding Social Networks in Urban Space - Computational Approaches for Urban Environments - page 311

Fig. 13.3 An individual activity space is represented as an ellipse (in black ). The ellipse captures
a user's most frequently used towers, shown as red circles ,wherelarger red circles indicate more
frequent visits

The value of the standard deviation is calculated for all x coordinates to obtain an
axis, and y coordinates to obtain a second, perpendicular axis. The ellipse is tilted in
a direction that captures the major axis (long edge) of the distribution (see Mitchell
2005 ).
As mentioned, the ellipse does not typically encompass all visited towers
and excludes those not frequently visited, to represent daily activity space (see
Fig. 13.3 ). It captures the essence of the central tendency, dispersion, and direction
of the user's travel patterns without including infrequent cell tower usage (such as a
traveler's phone call from the airport).

13.3.2 Assessing Overlap of Activity Spaces

After each individual is assigned an activity space, we quantify the similarity
between an ego's and an alter's activity spaces. A method for finding whether two
activity spaces are similar is not straightforward. The percent overlap between two
activity spaces will not account for how much physical area two friends' spaces
share. Additionally, using the area that two activity spaces share does not tell us
how big their spaces are, i.e., whether this shared area is actually “convenient”
relative to their whole activity spaces. Also, with these two methods, we will not