Millions of People's Location Data Revealed a 'Universal' Pattern In Study

More than half of people on Earth live in and around cities, and the population drift from rural to urban spaces is projected to increase in the coming decades. As cities continue to grow, it will become even more important to understand the complex social interactions occurring inside them, which influence everything from productivity to the spread of infectious diseases.

This need for new perspectives on urban bustle inspired researchers, led by MIT’s Senseable City Laboratory, to tap into data collected from millions of anonymized mobile phone users with the aim of filling a key gap in models of human movement in cities. 

The results reveal what the researchers describe as a new “universal visitation law of human mobility” that “opens up unprecedented possibilities” to predict flows between locations and that can be applied to cities as diverse as Boston, Singapore, and Dakar, according to a study published on Wednesday in Nature. 

The new research is “the result of years of research” at MIT, in collaboration with physicist Geoffrey West at the Santa Fe Institute, according to a joint email to VICE from study authors Carlo Ratti, director of Senseable City Lab (SCL); Paolo Santi, who leads the lab’s MIT/Fraunhofer Ambient Mobility initiative; and Lei Dong, a postdoctoral associate at SCL. 

The team’s approach was “motivated by the fact that current research focused on characterizing human flows between distant places” has “overlooked frequency of visitation,” Ratti, Santi, and Dong said. “We start by empirically characterizing this relationship and found that it could be described by a very clear mathematical law.”

In other words, the researchers aimed to combine established models of human mobility that have primarily fallen into two categories: large-scale maps that track aggregate numbers of people moving between locations, and more granular studies that track the frequency with which one individual visits different locations. 

The aggregate techniques, which include methods such as the gravity law and the radiation model, can provide insights about mass mobility in cities, but they typically don’t include information about recurring visits to locations by the same individuals. In contrast, techniques that focus a single person’s activities, like the exploration and preferential return (EPR) model, can shed light on the frequency with which an individual may visit key locations, but they tend not to provide a broader view of these repeated movements in a large population.

To bridge the divide between these models, the team amassed five different mobile phone datasets collected from 2006 to 2013—provided by Airsage, ORANGE/SONATEL, and Singtel—that “contain more than three billion time-stamped location records of more than eight million anonymized users,” according to the study. (Though mobile phone data is essential for these comprehensive academic studies, it’s worth noting that some datasets can be abused by governments and corporations even when anonymized.)

Phone users were located in seven cities representing four continents: Greater Boston, USA; Singapore; Dakar, Senegal; Abidjan, Ivory Coast; and the Portuguese cities of Lisbon, Porto and Braga. 

“Among the different data sets available at the lab and through our sponsors, we selected cities from different continents, as well as climates,” said Ratti, Santi, and Dong, in their joint email. “We also explored data at different geographical scales, ranging from metropolitan areas to entire countries.” 

Despite the dazzlingly distinct skylines, demographics, and characters of the studied cities, the researchers found that their residents and visitors all adhered to this universal visitation law, as described in the study: “the number of visitors to any location decreases as the inverse square of the product of their visiting frequency and travel distance.” 

In this way, the study exposes a strong mathematical pattern linking the likelihood of individuals visiting a location with their past movements and the distance they have to travel, regardless of the city these people may live in. Put differently, people frequently visit locations that are geographically closer to them. 

“The notion that distance and frequency of visitation are related is in accordance with intuition,” noted Ratti, Santi, and Dong. “What is surprising is that the relationship between these two quantities can be described by a simple and clean mathematical law.” 

“What is even more surprising is that the same law applies to different cities, and it applies also at larger macro-regional scales,” they continued. “Considering how different the Boston metropolitan area is from Ivory Coast in terms of culture, economy, climate, etc., it is really surprising to see that, when it comes to how people move in space, Bostonians and Ivorians behave essentially in the same way.”

In addition to revealing these basic human commonalities, the new approach can provide more accurate predictions about all kinds of urban exchanges and encounters. Indeed, the team is especially interested in understanding how their law factors into the spread of diseases, which is more topical than ever as the COVID-19 pandemic lurches its second year.

“We have used our inverse visitation law in combination with epidemic models, and shown that current containment policies mostly based on restricting distance of movements might be not effective, if frequency of visitation is not taken into account: a single visit to a far-away place might be contributing less to virus spreading than frequent visits to a nearby grocery store,” the three authors said. “Thus, effective policies should rather limit the distance and frequency of visits.”

The new law also provides empirical validation for established theories about human mobility such as the Central Place Theory, which suggests that people visit the closest possible location for their needs or wants, leading to distinct clusters within settlements. The study corroborates this theory for the first time with large-scale data, according to the authors. 

The team hopes to build on their findings by testing out other long-standing assumptions about urban mobility using their mathematical model.  

“A possible avenue for further study is to refine the analysis to establish, if existing, a connection between attractiveness of a location—i.e., how many visitors, and from what distance they are willing to travel to reach that location—and the point of interests, such as number and type of stores, theaters, schools, malls, etc.,” Ratti, Santi, and Dong said.  

“Such an analysis would further corroborate possible connections with existing urban theories, such as Central Place Theory,” they concluded.