For over two years, Mobility as a Service (MaaS) has been one of my primary focus areas, both personally and professionally. While we’ll define MaaS and discuss the role of big data in MaaS in the following sections, the importance of this concept grows each year for several reasons. These include increasing awareness of sustainability, challenging macroeconomic conditions in both developed and emerging countries, and strong user adoption of inter-mobility solutions.
For years, major map platforms like Google Maps, Yandex, TomTom, and others have leveraged mobility data to deliver value to users while monetizing insights through various revenue models. These include advertising and practical use-case solutions such as traffic optimization.
Recently, I came across an excellent book, Big Data and Mobility Services by Zhang, Song, and Shibasaki, along with insightful articles from academic literature. These inspired me to write a humble review of established methods and known applications, merging the insights from these sources with my own experiences. Unfortunately, publicly available datasets for mobility are scarce. However, with a little market research, you can create synthetic datasets that approximate real-world conditions and tailor them to your goals.
Before diving deeper, I encourage you to create an account on my blog so we can stay connected and exchange ideas.
What is Maas?
Mobility as a Service, or MaaS, is a fancy way of saying “you don’t need to own a car to get around.” It’s about bundling every mode of transportation—buses, trains, ride-hailing, bike shares—into one easy-to-use package. With a system , you can plan, book, and pay for a trip that might involve hopping between a train, a scooter, and a ride-share. Back then, we saw definitions from 90s like “intelligent information assistant for travel needs”. I think this definition will be more relevant in coming days, because it is not just about orchestrating mode of transportations, every day with companies like Uber, Grab, Thinkoff and so more they are entering to provide everything we may need during mobility.
But let’s not get lost in definitions. MaaS isn’t just a buzzword; it’s a growing ecosystem shaped by technology, economics, and evolving user behavior. While I’ll skip the detailed examples of MaaS here—that’s reserved for a future piece on MaaS growth strategies—this article will focus on its interplay with big data and how it transforms mobility as we know it. Instead, we’ll explore how players can utilize big data and established methods from the literature to process it effectively.
Methods
In the end this is an article, not a book and hard to cover every concept around in a single post. I am aiming to create separate sections for each area however today, we will try to understand what they are and what are known methods. I hope this article will be a good starting point for you.
Spatio-temperal data processing technologies
Spatio-temporal data processing is about weaving together the where and the when—analyzing data that’s tied to both location and time. Think of it as the GPS and the clock working hand-in-hand to uncover how people move through space. In MaaS, this means tracking vehicles in real time, analyzing traffic flows, and studying user travel patterns to understand their rhythms. These insights are the backbone of systems that adapt on the fly, offering smarter routes, balanced resources, and ultimately, a better travel experience for everyone.
One of the popular area of spatio-temporal analysis is Significant Place Analysis. Imagine tracing the rhythms of someone’s life—morning coffee at their favorite cafe, afternoons at the office, and weekend trips to the park. That’s what significant place analysis is all about: identifying locations that anchor users' routines. This technique uses background tracking to collect spatial and temporal data, observing when and how often people linger in specific spots. For example, analyzing time spent at locations during weekdays versus weekends or at different times of the day can reveal their home, workplace, or leisure destinations.
These insights are a goldmine for mobility services, supporting use cases like targeted advertising, resource allocation, and sustainable routing. Advanced AI models, like Hidden Markov Models and clustering algorithms, elevate the precision of this analysis by learning nuanced behavioral patterns from sprawling datasets.
Travel similarity estimations and clustering techniques
Travel similarity estimation is like decoding the shared rhythms of human movement. The goal is simple: group users with overlapping paths and behaviors to optimize mobility services. But the science behind it—from clustering techniques to advanced pattern recognition—is anything but.
Consider k-means clustering, a method that segments users based on travel behaviors, such as recurring routes or preferred transport modes. Even seemingly mismatched journeys can align through techniques like Dynamic Time Warping, which adjusts for variations in speed or timing to find deeper connections. These methods form the backbone of shared ride optimization or subscription plans tailored to user clusters.
Imagine a music festival—planners use historical travel data to predict and accommodate surges in demand. Shared ride services refine pickup routes, reducing detours while maximizing efficiency. This isn’t just about saving costs; it’s about creating a mobility experience that feels intuitive and personal, one cluster at a time
Data sources and their role in decision making
Data sources are the foundation of MaaS, acting as the arteries through which actionable insights flow. In a world of constant motion, GPS coordinates, IoT devices, transaction histories, and even weather data are the building blocks that reveal the true pulse of mobility systems. These datasets are not isolated entities; they merge into a rich tapestry that tells a story about how people move and what they need.
For example, integrating weather data can revolutionize decision-making. On a stormy day, travel patterns change—users may prefer ride-hailing over public transport, and traffic snarls grow worse. AI techniques like data fusion algorithms blend weather forecasts with GPS and traffic data to predict these shifts, enabling operators to adjust in real-time. Sustainable routing is another critical application. By combining road types, emission metrics, and external conditions, MaaS platforms can suggest eco-friendly routes, cutting emissions and supporting global sustainability goals. This is where advanced AI, such as knowledge graphs, steps in, connecting seemingly unrelated datasets to uncover hidden mobility patterns.
In practice, the power of data sources lies in their ability to inform dynamic systems. Social media feeds can detect delays, IoT sensors highlight bottlenecks, and transaction data predicts peak periods. Together, they create a mobility network that doesn’t just respond—it anticipates and adapts.
Mobility data optimization and dynamic pricing
Beneath the seamless user experience lies a complex dance of route planning, fleet allocation, and real-time adjustments. Dynamic pricing adds another layer, turning user demand into actionable strategies for balancing supply and profitability.
Picture this: an AI-driven system monitors a sudden surge of commuters during a music festival. Dynamic pricing algorithms kick in, using reinforcement learning to adjust fares in real-time, balancing availability with affordability. Meanwhile, fleet optimization models ensure that vehicles are strategically deployed, minimizing idle time and maximizing coverage.
Sustainability plays a starring role here. Carbon footprint estimation models allow platforms to incentivize greener choices. For instance, users choosing carpooling or low-emission routes could see reduced fares. Multi-agent systems simulate interactions between users, vehicles, and infrastructure, crafting a MaaS ecosystem where efficiency and sustainability go hand in hand.
The beauty of mobility data optimization lies in its adaptability. It transforms static systems into living, breathing networks that evolve with demand, ensuring that the promise of MaaS is delivered in every ride.
In the coming months, I’ll explore each of these sections in greater detail, diving into specific use cases and offering analytic reviews of the methods discussed. These deep dives will highlight how data scientists and mobility providers alike can harness these tools to shape the future of MaaS. Stay tuned for more insights and practical applications.
Comments