Embedded Files
Part 1: Analysis of surrounding Census tracts
We believed that including these population characteristics in our analysis would allow us to understand the demographics of the Census tracts that the 19, 20, and 51A lines run through. We also included tracts that lay within a quarter-mile walking buffer from each line in our analysis. We specifically decided to create maps pertaining to minorities within these tracts, such as percentage of Black and Asian populations and percentage of low-income households, as these were populations that were more likely to use public transit or depend on transit for their travels. We also mapped out the surrounding tracts by percentage of workers who take public transportation as we perceived this to be a vital trip purpose and one in which workers, specifically, could benefit from greater travel time predictability along their commute. Finally, we plotted the percentage of seniors by Census tract as this demographic group is less likely to use smartphones and would likely value having real-time information at bus stops they frequent.
Race/ethnicity
According to a study on transit ridership done by APTA, the proportions of Black and Asian people that ride public transportation is much higher compared to their makeup in the US population. These are likely transit-dependent riders that use transit for a variety of trip purposes, including commuting to work. Thus, they would likely benefit from more timely information on their trips.
The Bay Area, in specific, also has a large Asian population, with 27 percent of residents identifying as Asian American or Pacific Islander (AAPI).
In these maps, we find a high concentration of Black residents in Downtown Oakland, West Oakland, and around the Fruitvale neighborhood. Meanwhile, Asian residents are more evenly distributed across the analyzed Census tracts, with the highest concentrations near Downtown Oakland and the north and western parts of Alameda.
At the same time, the map of autocorrelation shows tracts with similar characteristics clustered in Downtown Oakaland for both percent Black and percent Asian residents. Census tracts in Fruitvale are more similar for percent Black residents, while tracts in Alameda are more alike for percent Asian residents.
Income and transit ridership
Low-income households may be less likely to have access to a smartphone or a data plan and may, therefore, rely on static schedules for bus arrival times. Installing real-time information displays in Census tracts with high percentages of low-income households could be especially beneficial for these residents as they may have no other means of accessing live bus arrival times.
Like the previous two maps, we see a high degree of similarity across census tracts in Downtown Oakland (autocorrelation), which extends to the Western portion of Alameda for transit ridership.
Age
Seniors are also less likely to use smartphones that would allow them access to real-time updates on bus arrival times. This is also a population group that may not have access to a personal vehicle or may not feel comfortable driving one. We plotted Census tracts by the population of seniors (adults over the age of 65) and found a high concentration of them in Downtown Oakland. In terms of autocorrelation, we see a high degree of similarity in Downtown Oakland (as well) and somewhat less similarity (though still significant) across Alameda and Rockridge.
Across all variables, spatial autocorrelation was statistically significant (p < 0.05) in Downtown Oakland and Fruitvale.
Smartphone and Internet access
While income can serve as a proxy for Internet access, the Census also provides data for Internet access by Census tract. As mentioned before, those without Internet access likely have no other means of obtaining real-time information for bus arrival times and would benefit greatly from real-time information displays.
As expected, many Census tracts with low Internet access also had low smartphone ownership. Interestingly, some tracts around Oakland have high Internet access but lower smartphone access. Riders in these tracts could benefit from real-time information displays to access arrival times outside of their homes.
From looking at these maps, it appears that Census tracts with greater concentrations of low income households and greater transit ridership do align with tracts that have more people of color. Census tracts with greater concentrations of seniors also seem to align with tracts with less smartphone access. Installing real-time information displays at bus stops in these tracts would likely be beneficial to multiple minority and age groups.
Part 2: Ridership
Ridership was another factor we considered in our analysis of which route and bus stops to implement real-time information displays. Initially, it appeared intuitive to prioritize installing displays at bus stops and routes with high ridership, but these routes have higher frequencies that translate to shorter wait times at their stops, reducing the benefit of such displays. Additionally, those without smartphones or Internet access who board at stops where many other transit riders might also be waiting may have the option to inquire others about bus arrival times compared to those at stops with only a few riders. Therefore, real-time information displays may actually be more beneficial at routes or stops with more unreliable service and fewer riders and might have the ability to attract riders to these stops or lines.
Boardings
We obtained daily average boarding data at each stop for Winter 2022 for each line and performed some summary statistics on this data by line below. We also split the summary statistics into weekdays, Saturdays, and Sundays and made boxplots for each line to better visualize the distribution of the data.
Stops along the 51A line clearly have the greatest daily average boardings compared to the other two lines. Additionally, the distribution of boardings at bus stops for all three lines seem to be skewed right, with plenty of outliers towards the right tail end of the distribution. Interestingly, line 20 has the greatest number of outliers, suggesting that although the average number of boardings across all of its stops may be low, there is a sizeable number of stops where the passenger boarding is relatively high.
Highest Ridership Census Tracts
We were then interested in taking a deeper dive into understanding the travel choices of those within Census tracts that had the highest number of public transportation users compared to all other tracts.
We find that the tracts that have the highest transit ridership also seem to have higher counts of walking and biking. We recognize that taking transit often involves active transportation for first and last mile access, so we decided to then evaluate walkability and reliability around stops with the highest ridership.
Walksheds around highest ridership tracts
Using the Summer 2022 weekday average boarding/alighting data provided by AC Transit and 511 API data, we created isochrone maps to show the walksheds of selected bus stops along each route that experienced high passenger ridership and/or excessive delays (greater than 5 minutes). The following maps show walksheds for three stops for AC Transit line 20 in oranges and three stops for line 51A in blues.
This bus stop sees around an average of 19 boardings on a summer weekday and experiences delays of 5 minutes or more.
This bus stop sees an average of 37 daily boardings on a summer weekday and experiences delays of 5 minutes or more.
This bus stop sees the heaviest use of an average of 55 passenger boardings per day during summer weekdays and experiences delays of 5 minutes or more.
This stop experiences delays and sees around an average of 35 passenger boardings per summer weekday.
This stop also experiences delays and sees around an average of 35 passenger boardings per summer weekday.
This stop is not necessarily delay-prone, but it sees a very high average of 112 passenger boardings per summer weekday.
These maps provide us with some contextual understanding of the surrounding areas of stops with high ridership and that experience delays. For instance, we know from these maps that the above bus stops are located in residential areas whose walksheds are also bisected by major freeways, as in the case of line 51A at all the stops mapped with blue walksheds above. This is unlike the bus stops on the line 20 that are not bisected by freeways. Furthermore, the maps for line 20 emphasize how Alameda is somewhat small as the entire island can be traversed widthwise on foot, which suggests potential for improving the first and last mile access for these bus stops. In turn, they could be better candidates for real-time displays, given that their walk sheds cover a larger unobstructed area.
Part 3: Performance and delays
We also took into consideration that stops where riders must wait for the bus for long periods of time would likely benefit from real-time information displays. To analyze which bus stops these were, we obtained real-time delay data from the 511 SF Bay API during peak and off-peak hours. We also plotted AC Transit's on-time performance data for the year of 2022. These gave us two perspectives into the performance of each line.
We decided to map out the bus stops within each of the three lines with average delays that exceeded AC Transit’s on-time performance threshold of five minutes. We hoped to create these maps to help us decide which stops within the most unreliable line would benefit from real-time information displays.
To view and interact with these maps, please click on the titles of each map below.
We also looked towards AC Transit's historical data on on-time performance for guidance. We were able to plot the 2022 on-time performance for the three lines and found that the line with the shortest headways, line 51A, has the most delays, while the line with the longest headways, line 19, has the least.
While the 51A might have the poorest on-time performance, the short headways also mean that a passenger waiting at a bus stop might have to wait up to 15 minutes, but on average around 7 or 8 minutes, to ride the bus. As a result, it might be more beneficial to have the real-time information at the stops of the 19 or 20 lines as a delayed bus can have a much more significant impact on a riders' wait time and total commute time.
Part 4: Accessibility
To determine accessibility to important destinations, we looked at community anchor institutions in the surrounding area. Specifically, we considered distance of bus stops to the nearest child care, healthcare, and elderly care centers for each line, which we deemed as necessities for many families. This data came from the California Public Utilities Commission.
Distance to child care
We found that the nearest child care locations were much farther for stops along line 20 compared to the other lines, although line 51A did have variability in distance to child care throughout at its stops.
Distance to healthcare
All three lines have significant variability in the distance to nearest healthcare, especially stops in Alameda and those in the Rockridge neighborhood of Oakland. The stops in Alameda, particularly for line 20, are quite far from healthcare facilities, up to 1910 meters away.
Distance to elderly care
All bus stops in Alameda have larger distances to elderly care, across all three lines. This indicates that the island of Alameda has very few elderly care facilities.
The above maps emphasize how the island of Alameda has fewer care facilities compared to its surrounding areas. It is also important to note that the distance calculated is straight-line or Euclidean distance between these stops and the facilities. As Alameda is an island and only has limited entrances and exits (six bridges and two tunnels), the actual distance between these stops and the facilities is likely greater than the straight-line distances shown here.
Header image from Flickr: Sullivan, P. AC Transit 1213 HT. CC BY-ND 2.0
Page updated
Report abuse