It seems that public transport demand is on the rise, which can be a good thing, except for left-behind passengers.
In mid-September 2021, a group of special needs students of Suffolk Rural College in the UK were allegedly left behind by their low-frequency school bus without a clear reason. This is sadly not an isolated case.
In October 2020, passengers were being left behind by bus services in Dublin. In response, Anne Graham the CEO of Ireland’s National Transport Authority (NTA) told RTÉ’s Morning Ireland that some passengers were left behind due to the limited capacity of bus services during the morning rush hour.
She also urged people to only use public transport for “essential trips” to lower the occupancy level. I would normally agree with her. But I also wonder whether congestion is THE CAUSE or whether there’s more to the problem that needs to be addressed on the supply side. Maybe a quick glance at some research literature can answer this.
- 5 most easily found studies about left-behind passengers
- 1. Inferring Left Behind Passengers in Congested Metro Systems from Automated Data
- 2. Estimation of Passengers Left Behind by Trains in High-Frequency Transit Service Operating Near Capacity
- 3. Passenger detection through video processing and signal sensors in the Boston subway to address left-behind passengers
- 4. Measuring Left-Behinds on Subway
- 5. Estimation of left behind subway passengers through archived data and video image processing
- So, how can left-behind passengers improve public transport?
- Related Posts
5 most easily found studies about left-behind passengers
Here are the 5 studies related to using data to count left-behind passengers to figure out ways to improve service performance.
by Zhu, Koutsopoulos and Wilson (2017)
- Two probability models, namely maximum likelihood estimation and Bayesian inference methods, fed with automated data were used to estimate and distribute the probability of left-behind passengers with high accuracy.
2. Estimation of Passengers Left Behind by Trains in High-Frequency Transit Service Operating Near Capacity
by Eli Miller, Gabriel E. Sánchez-Martínez and Neema Nassir (2018)
- This paper used a regression model known as a bi-level regression model to estimate the number of left-behind passengers on a high-frequency train platform in Chicago, USA.
- The model was fed the number of passenger arrivals between train stops (using smart card data and vehicle location data) to compare with manual counts of left-behind passengers and calculate the cumulative capacity shortage.
- The cumulative capacity shortage was said to be related to the number of left-behind passengers.
3. Passenger detection through video processing and signal sensors in the Boston subway to address left-behind passengers
by Andronikos Keklikoglou of Umass Amherst (2018)
- Finding the best way to count left-behind passengers at 2 train stations in Boston since the number of left-behind passengers was said to be directly related to public transport performance measures such as ridership, service quality and travel time.
- Surveillance camera-based object detection software and, Bluetooth and Wi-Fi sensing were used to count left-behind passengers.
- The APC data sets were cross-checked with the manual counts to determine which APC tech was more accurate and precise.
also by Keklikoglou but with Sipetas and Gonzales (May 2018)
- Showing ways to measure or estimate the number of left-behind passengers and distribute the left-behind passengers’ waiting times.
- Left-behind passengers can be counted using automated passenger counting (APC) via surveillance camera footage, detected Bluetooth and Wi-Fi connected devices and vehicle location data.
- 90% accuracy for estimating at least 1 left-behind passenger.
- Within 10% accuracy for estimating the total number of left-behind passengers during rush hour.
again by Keklikoglou, Sipetas and Gonzales (2020)
- Estimating the number of left-behind passengers (also in Boston) using existing data sources with new object detection software.
- Only smart card data was collected for entries into subway trains since the trains didn’t have automatic passenger counters (at least, at the time of research).
- However, surveillance camera-based object detection software could be used to count the left-behind passengers on the platforms.
- The APC data and train operations data were fed to the logistics regression models (the ones with the S-curve) and compared with manual counts on a normal operations weekday.
- Putting together APC data and train operations data worked for estimating the number of left-behind passengers within 10% accuracy.
So, how can left-behind passengers improve public transport?
From what I can deduce about the studies above, data about passenger numbers (using the people counting methods with high accuracies), service capacities, service frequencies, schedules and so on can be fed to probability models to estimate the number and probability of passengers getting left behind.
Seeing that the necessary data isn’t limited to people counting, it’s safe to say that there’s more to the problem than just more people showing up to the stations. It also shows that public transport operators and authorities are just as responsible for handling this issue on the supply side.
Using the data gathered and estimations computed, the operators and authorities can find ways to optimise their service frequencies, capacities and schedules to accommodate the actual number of passengers at different times of the day every week.
That’s why the first step in making sure that
ohana means family and family means nobody gets left behind is acknowledging that the operators and authorities are here to protect and to serve passengers’ needs. And that punctuality has a bigger impact on occupancy than occupancy does on punctuality.