Experiments to determine bus stop closeness threshold value
Below, we explain how to obtain the bus stop closeness threshold
value. For experiments, a traveler carried a mobile device and
boarded a CTA bus. We then measured the Euclidian distance to
the closest bus stop from the traveler’s device. From over 450 GPS
sensor reports, we plot the graph of Figure 5. The vertical axis
represents the Euclidian distance to the closest bus stop. The
horizontal axis represents the corresponding GPS sensor
report number. From Figure 5, we observe how the mobile device’s
Euclidian distance to the closest bus stop fluctuates during the
travel. When the traveler on a bus is at the bus stop, the distance is
at a minimum. As the bus moves away from the bus stop, the
distance increases. It peaks at the midpoint between two bus stops.
Afterward, it decreases and reaches a new minimum when the bus
reaches the next bus stop.
When the traveler is traveling via bus mode and at a bus stop, the
Euclidian distance to the closest bus stop is less than 13 meters.
Thus, for bus stop closeness threshold, we used a value of 13m.
For BSCR, we compute the number of times the Euclidian distance
to the closest goes below the bus stop closeness threshold per unit
time. We believe that if a traveler is traveling by bus, the BSCR
should be greater than if they are not travelling by bus. We also
evaluate the effectiveness of BSCR on predicting the transportation
mode in the proposed work.