Part 2: Intelligent Transit Planning & Advanced Transportation Analytics
In part 1 of Intelligent Transit Planning and Advanced Transportation Analytics, we focus on the importance of Multimodal Journey Mapping to understand the end-to-end commuter’s journey and how a robust Demand-Responsive System can revitalize the industry by providing more options to the end-users. In part 2, we extend the discussions on the benefits of building economically efficient networks and the use of descriptive and predictive insights in solving the worldwide problem of traffic congestion. For the ultimate success in the battle, building and strengthening data aggregation and exchange across organizations should be our ongoing priority. As a bonus point in the end, we share a few startup ideas on how to solve the most pressing problems in this city.
Economically Efficient Networks
The use of granular GPS data and advanced analytics enables cities to create a more cost-efficient transit system. Private companies like Uber and Lyft have recognized the opportunity of using GPS data to facilitate ride-hailing and sharing. They also use advanced analytics to power their dynamic pricing models. The combination has created trips with a lower total average cost compared to car ownership or traditional taxis. Public agencies are now taking a similar approach to create transit systems that are economically viable while still providing flexibility and availability for travelers.
Building a modern transit system can create real cost-savings for a city. Particularly for cities with heavy single occupancy vehicle congestion, the stakes are high. Studies by a European analytics firm, Inrix, identified the following:
In the UK, the worst traffic hotspots in 21 cities were identified and the cost to drivers of time wasted in congestion could amount to £61.8 billion over the next 10 years.
London had more traffic hotspots than any other city analyzed, and time wasted in gridlock at these locations could cost Londoners £42 billion by 2025
In Germany, INRIX Roadway Analytics identified the worst traffic hotspots in 27 cities and the economic cost to drivers could amount to €47.6 billion by 2025
The A7 in Hamburg is home to Europe’s worst traffic hotspot and time wasted in gridlock at this location could cost drivers €1.3 billion over the next decade.
Beyond the reduction in congestion-related costs, modern transit systems are economical for cities in several ways. By leveraging both GPS data and advanced analytics, cities can create sustainable and dynamic systems. Five applications of this approach are listed below.
Dynamic pricing models based on a multimodal trip to increase average trip revenue
Decreased transit infrastructure maintenance costs from predictive models
Improved demand and capacity planning to maximize revenue and minimize costs
On-demand transportation systems to reduce under-utilized transit routes
Partnering with private companies for end-to-end journey solution to reduce reliance on public infrastructure
The last point can be particularly beneficial for cities. Rather than huge public investments in projects that will take several years to be completed, cities are benefiting from the proliferation of private investment in transportation services. As an example, existing public transit infrastructure is often supplemented by private ride-hailing systems today. The image below shows the San Francisco journeys, estimated by Uber, where it’s likely that travelers used two Uber trips connected by Caltrain over a 24-hour period. Cities like San Francisco can adapt to changing transit needs quickly because of private investment that uses GPS data and advanced analytics.
The illustration below shows how Uber trips in the Bay Area are mostly short distance and widely dispersed. This portion of a traveler’s journey would be difficult to fulfill with traditional public transit infrastructure in an economical way. Similar to supply chain logistics, the last few kilometers of the journey is generally the most expensive. Private companies have built a symbiotic relationship with public infrastructure to manage this portion of the journey in innovative ways powered by GPS data and analytics. In the future companies like Lyft, Uber, Bird, Dropbike and Bridj will continue to play a crucial role in a multimodal transit system.
Descriptive and Predictive Insights
Amongst the advanced analytical methodologies being used to support the optimization of urban mobility (see table below), predictive and machine learning models can perhaps deliver the greatest amount of value. Through the study of daily or hourly population flow across various transportation alternatives from point A to point B (e.g. travelling from home to work), underpinned by data on factors contributing to efficient or inefficient flow of traffic (e.g. rain, construction, summer holidays, paydays, etc.) city officials and private entities in the transportation space can more confidently assign resources to absorb such changes and communicate accordingly to the public.
Predictive models have the potential to lower costs in the long term. For example, through lowering the number of buses servicing a particular route during the summer months at specific times of the day or by giving commuters a forward-looking view into their evening transit plan, to offer better and more affordable transit alternatives based on forecasts and simulations in real-time. The possibilities are endless.
Private companies are doing this to serve some segments of the population. INRIX, an analytics firm, and BMW, a luxury car maker, developed the industry’s first-ever in-car EV Range Finder + Intermodal Navigation system, giving drivers an accurate estimation of how far they can travel in all directions from their current location based on their current battery charge.
In the public space, Machine Learning techniques are being applied to improve large complex traffic systems. By using sensor arrays and computer vision helping to detect multiple street users and inform decisions on traffic flow. This sort of tools are making intersections safer (i.e. reduce traffic accidents).
Tech-enabled, responsive signals could help slow down vehicle traffic if average car speeds at an intersection exceed a certain threshold at which collisions with pedestrians become more deadly. New alert, in-car systems could also potentially give drivers a direct and quick warning of pedestrians and cyclists in the area.
Using live public data feeds from trains and buses, the Swiss-German technology firm GeOps and the University of Freiburg have launched an interactive map of the world’s major transit systems to show how major cities move. With the city of Toronto being depicted in Figure 1 below, transportation officials can use this real-time data to identify the most common problems in the network and build predictive models to forecast potential issues in the future to specify solutions to improve the flow of people across the network.
The figure above shows all public transportation agencies in the Toronto area. Each colour signifies one different transport line and each node represents one vehicle traveling in that line or route. The number of dots increases during peak hours of the day and is reduced or eliminated during off-peak hours.
In the public space, it will be critical to connect as many data points as possible from a multi-modal transportation system in order to adapt resources to expected patterns and give confidence in capital requirements to develop the right infrastructure in the right areas of a city.
Connected Data Across Organizations
The success of a modern transit system depends heavily on the ability to link locational device data together and map out complete transit journeys across multiple modes of transportation. The best case scenario would be to have a smartphone application that oversees access to any type of transportation system, public or private, or at the very least a combination of a physical smart card (e.g.. Presto) and a digital wallet application (e.g Transit) to aggregate individual mobility patterns under a unified user ID.
The technology to do this exists today. With the majority of travelers owning smartphones, particularly in urban areas, there is a plethora of location-based data available. Smartphones provide persistent device IDs, IDFA for iOS and Advertiser ID for Android, which can be captured through mobile apps or tracking beacons. These IDs provide a common join key across private companies and public agencies that can be used to unify data sets. This will require cooperation between different organizations however examples of this are already available.
In 2017, Alphabet subsidiary Waze entered into a data sharing agreement with the City of Toronto. The agreement is mutually beneficial. It provides Waze and its users with advanced notification of accidents, road closures, and other potential traffic disruptions. It provides the City of Toronto with individual movement data to better measure travel patterns and plan city streets accordingly. The future of transit planning will require more of these agreements to unify data sets across organizations.
Even without the presence of a smartphone device ID join key, it is possible to connect data sets across organizations. A joint study across universities in China and Singapore was able to join data sets from separate train and bike share agencies in one city.  The researchers built an association rule based method to match travelers’ train metro cards with their bike share cards. The model was able to match 573 pairs of cards with 100% accuracy in the sample data sets. By applying analytical techniques to join data sets across organizations, deeper insights can be gained for transit planning. Many of the future advancements for transit systems will depend on the interconnectedness of data across all stakeholders.
The application of analytical methods to optimize transit systems around the world is gaining more popularity throughout the world. Our research illustrates a consistent pattern across cities in various countries where more robust mobility solutions are being developed, with a strong focus on the connectivity of transportation systems and mapping population flows.
One such successful example is of the ITS model in South Korea, which not only has been successful for their country but is now also being purchased by many countries to implement.
Relatively newer transportation solutions such as Uber have strategically taken a large share of the market with the help of GPS enabled service for easier customer experience, but they continue to compete with other established transit systems rather than working with them.
The evolution of city transit systems needs to progress towards a complete commuting solution for the traveler, providing an end to end cost structure and an efficient connection between different platforms. These systems should be connected via a single profile, transmitting data to and from the individual’s profile, back to transit authorities for real-time and predictive decisions.
In terms of startup ideas, we believe the value-add is to solve some of the challenges still being faced today in these two key areas:
One: Impacts of high pedestrian density around major street intersections: Measure and quantify the direct and indirect impacts that pedestrian traffic has on other transit routes, namely for vehicle movement in downtown cores where peak hour pedestrian traffic can lengthen vehicle turning time and create a backlog of vehicles trying to access main arterial highways.
A good example is the complexity arising around Union Station in downtown Toronto where hundreds of people are walking to catch their trains while others are trying to leave the parking garages to access the Gardiner Expressway or Lakeshore Boulevard routes. There is further opportunity from analyzing the same impact given pedestrian traffic underneath the Gardiner Expressway, as vehicles need to wait for people to cross Lakeshore Boulevard, creating a further backlog.
Solutions to this include restricting pedestrian crossing at a major intersection only on the north side of the street during the afternoon peak hours and limiting the number of pedestrian routes (i.e. only York and Yonge streets) during peak hours, to access the waterfront or building pedestrian tunnels.
Two: Alternative options to avoid being stuck in messy commutes: By leveraging expected and forecast commute travel times, we would build a set of alternative options for travellers to avoid long wait times. This could be in the way of giving them tailored shopping, entertainment or restaurant experiences in order to minimize the amount of time spent idle and maximizing their time on other activities that are either required or helpful for a balanced lifestyle.
For instance, if a regular commute is going to be 30 minutes longer by leaving now, our company would provide different offers from partnering businesses to incentivize people to stay around downtown longer until the expected commute time dies down. This would require entering into partnerships with Restaurants, Shops, Event Planners etc. From this information, an even more complete customer 360 view can be achieved yielding opportunities to sell this data to other parties at an aggregate level with sentiment analysis and population segmentation to drive better offers in the long term.
A complete transit solution will give ease to the travellers and will reduce the temptation of them utilizing an entirely different mode of transportation, such as driving. This can also reduce the number of cars currently on the roads today as the daily drivers will reconsider the possibility of using transit, given our ability to achieve a connect efficient structure.
Are you based in Toronto and working on Internet of Things - IoT initiatives?
Talk to us and share your point of view! Together we can help shape the future of our communities and contribute to Canada's leading role in Artificial Intelligence.
About the Author(s)
This article was originally written by Elie Yachoui and his Master of Management Analytics teammates at Queen's University’s Smith School of Business as a blog in their Operations and Supply Chain Analytics course taught by Prof. Anton Ovchinnikov.
Editor: Michelle Liu