Unlike many modern cities, say Los Angeles or Washington, D.C., Vancouver does not have a massive highway system that leads from the suburbs to the city's centre. This was a deliberate choice by the urban planners of Vancouver to keep the city from being a commuter town. Given its reputation for having a vibrant urban environment that is also pedestrian friendly, the planners have clearly achieved their goals. However, in not having a massive highway system, the residents of metro Vancouver rely heavily on the city's public transit. While Vancouver has one of the best public transit system in Canada, it has its shortcomings. Bus pass-ups and bus bunching are two problems that frequently occur in Vancouver.
A bus pass-up occurs when there are people waiting at a bus stop but a bus cannot stop to pick them up because it is already full. Typically, this problem affects riders who get on the bus later in the route. During rush hours, some riders could be passed up multiple times before a bus could pick them.
As a graduate student at Simon Fraser University, I know this problem well. Like many students of the university, I rely on the public transit to get to school every day. Whenever a pass-up occurs, it is always a frustrating experience. This problem is such a source of concern that SFU has a dedicated forum for students to voice their concerns. However, this forum is not well known, so very few data points have been collected.
Because of this frustration, I decided to study this problem as part of a class project. To begin, I collected data from TransLink's open API over several weeks. This API is an interface for computer programs to collect data from TransLink's system. I began with some basic statistical analysis on the data and I found several surprising facts. On average, the bus trips tend to finish on time or faster than the times listed on the schedule. Furthermore, there are usually more buses on the road than the number of buses scheduled as well. These two results are also more apparent during rush hours on weekdays than other times.
Based on these results, I believe that TransLink does not strictly adhere to the published schedule. Particularly, when there is high demand, there are more buses and they run slightly faster than schedule to satisfy this demand. As a rider, I used to think that buses tend to run slow and they are never on time. Given the facts shown by the data, I have to think that I was biased. Buses probably seemed slow because I was in a hurry. Evidence suggest that buses schedules are adjusted dynamically to satisfy demand. Without this knowledge, it is easy to assume they are late when the arrival times do not match the published schedules. Also, it is comforting to know that there is usually more service than advertised.
Aside from running time, another source of frustrations for riders come from bus bunching. As any frequent user of public transit will tell you, on a busy route, sometimes one would wait much longer than the expected wait time and only to see two or more buses on the same route showing up at the same time. This phenomenon is called bus bunching. When buses are bunched together, the earlier bus tend to have more passengers and lead to uncomfortable conditions for the riders. On the other hand, capacity of the latter bus is usually not filled to capacity. Therefore, this is an inefficient use of resources on the road.
Through visualization of the collected data, we can clearly see this phenomenon in Figure 1. Each line in the figure represents one bus trip. In this figure, the vertical axis represents distance and the horizontal axis represents time. Whenever two lines intersect or become close to each other, we can infer that two buses are close together on the road and a bunching has occurred. By comparing the two figures, we can see that bunching primarily occur during weekday morning rush hours (Figure 1 (a)). On the other hand, buses simply do not catch up to each other on weekends (Figure1 (b)). While not shown in the figures, bunching also occurs during evening rush hours.
Bus trajectories plots for 135 East.
(a): Morning rush hours on 2015-10-20 (Tuesday).
(b): Morning rush hours on 2015-10-25 (Saturday).
In researching on the problems of bus bunching and bus pass-up, I found that most cities share these problems (Please see Improving mass transit operations by using AVL-based systems: A survey by Moreira-Matias et al. for examples and research on public transit). There are many proposed solutions such as creating express routes, changing locations of the bus stops or increasing road capacity. While implementing these ideas will alleviate some of the concerns, these problems will continue to persist as the city continues to grow. For long term solutions, the city must investigate and research this problem. With the newly installed Compass card system and the existing automatic vehicle location data, there is an unprecedented amount of information available to understand traffic patterns within the city. As shown through this brief report, data analysis can directly reveal patterns and predict future transit problems. We must apply machine learning techniques to the data to predict the needs of the city. Urban planners of Vancouver can use the results of such studies to shape this city in a sustainable way.