Abstract
|
|
In this paper, we use a webcam observing an intersection in midtown Manhattan to study the aggressiveness of New York cab drivers. After developing an operational definition of “aggressive”, a manual analysis of one week of data demonstrated that cab drivers are statistically more aggressive than non-cab drivers.
|
Introduction
|
|
New York taxi-drivers are famous for their aggressive driving habits. Via casual observation, taxi-drivers tend to position themselves as closely as possible to the stopping line when the traffic signal demands a stop. Presumably this allows taxis to get the fastest possible start when the light turns green. This webcam provides an ideal set of images to confirm or refute this observation.
|
Methods
|
|
A week of images were sampled from the webcam and analyzed manually. Cab driver aggressiveness is calculated for each image by dividing the proportion of cabs on the front line from the proportion of cabs in the scrutinized area. The same analysis is repeated for non-cab drivers. Finally, a T-test is used to determine whether the two groups differ statistically.
Above is a depiction of the Scrutinized area. Vehicles must appear within this area to be counted. Vehicles must appear before the green line to be considered "on the front line".
|
Results
|
|
The following graph illustrates the distribution of the collected data. As you can see, the probability mass of the cab distribution is primarily to the right (more aggressive) than that of the non-cab distribution. Interestingly, the variance of the cab data was much larger than that of the non-cab data. Finally, a T-test demonstrated that with certainty greater than 99.9%, the two groups are statistically different.
|
Privacy
|
|
The resolution of this camera is not sufficient to read license plates or taxicab id numbers, so the risk of identifying an individual via webcam images is very small. As a result, it is unlikely that anyone might experience a loss of privacy as a result of this webcam or this research.
|
References
|
|
Camera: https://www.riotmanhattan.com/riotmanhattan/webcam.html
Project Paper (PDF)
|
