ProRail, the appointed railway infrastructure manager in the Netherlands, has partnered with Xomnia to harness the power of anonymous big data to understand crowds and better plan and develop its platforms. As the infrastructure manager expects a 30% increase in the total demand for passenger and freight transport by 2040, it wants to ensure that its decisions to adapt its facilities to this increase are informed by data.
To achieve this, our machine learning engineers Yu Ri Tan and Jan Scholten are creating an environment that allows ProRail to analyze anonymous crowd flow data in a fast, tested, scalable and reproducible way. This environment consists of several services on the Azure cloud, as well as tailor-made software packages.
Using these infrastructural tools, ProRail will be better able to use the anonymous data that it has been collecting for years to plan its infrastructure developments*.
“It's great to work with Yu Ri and Jan. We now have a productionized and scalable crowd flow toolbox, integrated in our Azure environment, which allows our team to work with large amounts of data” --- Frank van Schadewijk, Product Owner
For years, ProRail has been collecting anonymous data about crowds on its platforms, recording traffic and its flow in different stations across the Netherlands. The anonymous data is collected using sensors, and is delivered as an output of tabular information.
ProRail aims to use this anonymous data to better understand the human behavior on its platforms, and hence plan it in the best way to guarantee safety, comfort and efficiency. This is especially important as the infrastructure manager expects a 30% increase in the total demand for passenger and freight transport by 2040, and wants to be as prepared as possible to accommodate this increasing demand.
The infrastructure manager has already launched data-driven initiatives to enhance its operations using the sensor data it has collected anonymously. However, due to the variety of business questions ProRail wants to answer using this data, it needed tooling to help it process and analyze its anonymus big data quickly.
To achieve this, ProRail approached Xomnia to create the fundamentals necessary to make useful analyses of its anonymous data, now and in the future. Therefore, over the past year, our machine learning engineers have been working to develop the data transformation, analysis building blocks and cloud infrastructure that form the architecture that ProRail can use in processing billions of rows of data per day.
Our machine learning engineers are developing the toolbox of building blocks and cloud infrastructure hand in hand with a number of business questions, to make sure it complies with ProRails’ needs. The tools include the software to do data analyses, but they also apply to the infrastructure backing the tools, such as distributed computing solutions that can process large amounts of data.
Using the building blocks, it becomes much more convenient, reproducible and faster to do certain analyses, which can be used for the different tasks that the infrastructure manager is legally obliged to perform*. Moreover, the toolbox is used to monitor the quality and consistency of crowd flow data, so that results are reliable and comparable.
Applications of the toolbox include creating line counts, calculating the number of people in a certain zone, and collecting trajectory data about the flow of people on the platform from a certain region to another. Data up till one day earlier can be processed, providing a view on the crowd levels on a day-to-day basis.
Using the toolbox, ProRail can also easily determine the locations of the doors and the numbers of people coming in and out of each one. This can help in understanding how the topology of a platform or other factors affect the efficiency with which travelers disembark and embark a train that arrives into a station. Using this knowledge, ProRail can take decisions that improve punctuality.
In addition, the toolbox’s output is being used by a team at ProRail that is specialized in explaining delays. While this team uses data coming from different databases, the team could not previously process nor interpret data collected during the train’s stop. Using the toolbox, this data can be analyzed and passed to this team, which helps them better understand the reasons for delays.
The toolbox of transformation and analysis building blocks is owned by ProRail’s crowd flow team and is ready to conduct many ad hoc analyses of anonymous big data. The three pillars guiding the development of the toolbox are travelers’ safety and comfort, as well as efficiency in operations.
Using the toolbox, for instance, decisions such as making platforms wider or relocating stairs to enhance the flow of passengers can be made in a more informed manner based on data rather than speculations. It can also be used to help in explaining reasons for delays better, take decisions that are informed by data when building new stations, and many other applications.
* Based on the Railways Act and the Passenger Rights Regulation, ProRail has a statutory duty to ensure safe transfer facilities for travelers, etc., to take appropriate measures to guarantee the personal safety of travelers at the transfer facilities, and to manage the risks.