Extracting urban activities through aggregate cellphone usage
Authors
Abstract
Urban areas have been facing a rise in carownership and commuting that lead to increased
congestion and pollution, and lose of space for
productive activities. To address these problems,
urban planners should develop means to profile
urban activities in a dynamic way. However,
obtaining data to create the required information
with traditional survey methods is expensive and
time consuming. Meanwhile, cellular networks
produce massive amount of data that could allow
us to sense the presence and movement of people.
This study applies passive mobile positioning
data such as, Call Volume, Handover, and Erlang
to detect the spatiotemporal distribution of
activities. Our observations are based on hourly
aggregated cellphone data obtained from a
dataset of communications in Lisbon, Portugal.
Fuzzy c-mean clustering algorithm was applied
to the cellphone data to create clusters of
locations with similar features in what respects to
two aspects of activities: daily patterns and
intensity. In order to validate those clusters as
actual predictors of human activity we compare
them with clusters formed using ground truth
variables: presence of people, buildings, POIs,
bus and taxi movement. In what respects to
identifying daily patterns of activities, the Erlang
data provided a better match with the ground
truth giving 68% of accurate predictions. In the case of the intensity of activities the Call Volume
data provided the highest match with the ground
truth yielding 79% of accurate predictions.
Hence, results demonstrate the potential of
aggregate cellphone data in detecting density of
activities that are superimposed on the different
activity patterns, which is fundamental piece of
information for transportation and urban studies.