DATA
Our queuing model described in Section IV uses two
pieces of data as input. 1) The rate of arriving taxis at
each terminal and 2) the number of passengers that arrive
at each terminal’s taxi stand. In the simulation that we
have developed, we obtain the first from the ChangiNOW
system when taxi drivers indicate their intention to head
to the airport and the second from historical flight arrival
data. Our dataset consists of one month of taxi journeys in
Singapore. The dataset we used contains millions of taxi
records, where each record contains the time-stamp, GPS
coordinates, driver number, etc. as well as the operational
status of the taxi. Records are logged at short intervals and
allow us to track taxi journeys over the course of the month.
The flight manifest data provides us with the flight id, the
number of passengers arriving on each flight and the actual
time the flight landed. By cross-referencing the flight ids
with airline schedule data available online, we were able to
determine the terminal at which the flight landed.