Sections

Services

Information

UK United Kingdom

Your phone knows the three places you visit each day

We lead busy, complex lives. But how many different places will you visit today? And how many different ways could you organise your travel between those places? The answer, according to a new study published…

Your daily movements are simple, predictable and useful to know. Thuany Gabriela

We lead busy, complex lives. But how many different places will you visit today? And how many different ways could you organise your travel between those places?

The answer, according to a new study published in the Journal of the Royal Society Interface and using, among other measures, mobile phone data, is: not many!

An international collaboration led by MIT shows more than 90% of people visit fewer than seven distinct places each day. And just 17 basic movement “motifs” (individual daily movement networks) are sufficient to describe 90% of our daily movement patterns.

These new results add to our understanding of the remarkable patterns that underlie human movement. Previous studies have already found strong patterns in larger-scale human movement over months or years, such as through analysis of the movement of $1 bills in the US.

This latest study, however, addresses our smaller-scale, workaday movement.

Mapping movement patterns

The study’s results can be summarised in three main findings:

1) People visit remarkably few places in the course of their day. The average number of places visited was approximately three, with few people visiting more than six places in a day.

2) The paths people take between those few places are also highly constrained. The 17 movement “motifs” found in 90% of the daily movement can themselves be described using just four rules. These rules essentially describe the tendency of people to perform “tours” (visit several places in one round-trip) rather than multiple individual trips between places.

3) People vary little in their movement patterns over time. You are most likely to visit the same number of places and using the same movement “motif” tomorrow as you did today (even if the actual places you visit and routes you take are different).

Behind all these regularities is the fact that space constrains movement. When we take into account where we must go (to work, to home, to bed), we have many fewer options for where we could go than one might think.

Back in the 1950s, the geographer Torsten Hägerstrand succinctly described this link between movement in space and movement in time, a topic now known as “time geography”.

Today, we have the capability to generate much greater volumes of data than could have been imagined in the 1950s (the so-called “Big Data”).

It is perhaps surprising, then, that two of the three data sets used for this study were collected using a technique that would have been familiar to Hägerstrand: travel and activity surveys.

In a travel survey, volunteer participants record their travel inventory over the course of a few days. Although the two surveys used in this work are large by travel survey standards, involving thousands of participants in Chicago and Paris, they are inevitably limited in their scope and size, being expensive and laborious to collect. In short, these two data sets are from the era of “Small Data”.

Location tracking using mobile phones

The third data set used is the one likely to draw most attention, as it relies on data about the locations of Parisians’ mobile phones. As everyone with a mobile phone should know, in order to be able to route calls to you, your phone company must track the location of your phone whenever it is turned on.

But this data does not relate to your exact location, only to the location of your nearest cell phone tower. In most urban areas, that will typically be 200 metres or less away from your actual location.

the|G|™

In order to use the phone data, however, the researchers needed to apply considerable data cleaning and pre-processing. Only data about the cell locations of phone calls and SMS messages are typically available to researchers, and as a result only the most active phone users were included in the study.

To generate the patterns observed in the results, many small-scale movements between cells had to be ignored.

For example, repeated movements between neighbouring network cells (such as can occur between two ends of a building) were filtered out of the data.

Indeed, while this study does give us new insights into a smaller scale of human movement than previous work, such studies cannot (yet) tell us about the patterns of human movement between nearby buildings, within a single building, or even within rooms.

Such micro-scale information can be important: the distance between the dock and the jury box in a courtroom may be small, but it is significant.

What does this mean, then?

The results of the Interface paper have implications for many applications, such as planning our cities and transport systems, and controlling the spread of infectious disease outbreaks.

JR_Paris

But they also have implications for our personal location privacy. Knowing that the range of human mobility patterns is so small and strongly correlated does potentially make it much easier to predict a person’s future mobility patterns.

So what should you do if you want to make your own daily patterns harder to predict? According to the study results, you could try to visit more locations each day, in particular aiming to fit more than six distinct places into your daily itinerary.

More importantly, you might try varying the order in which you visit places day-to-day, changing the routine regularly.

But our complex and ordered lives rarely provide opportunities to, say, visit the pub for a quick drink after dropping the kids off at school in the morning, instead of going in the evening after work; or to add multiple back-and-forth trips to the gym during the work day.

For most of us, being stuck with complex lives means being stuck with simple and predictable movement patterns.

Join the conversation

8 Comments sorted by

  1. Wil B

    B.Sc, GDipAppSci, MEnvSc, Environmental Planner

    "So what should you do if you want to make your own daily patterns harder to predict? "

    "For most of us, being stuck with complex lives means being stuck with simple and predictable movement patterns."

    It's a mildly interesting article, but I'm really struggling to see why you needed to put some weird value/emotion stuff in it at the end.

    I don't know about you, but the risk of me getting killed in a car bomb, or tailed by a private detective, is rather low. So what and who cares if I have predictable movements. Why is this a bad thing???

    report
    1. Andrew McNicol

      PhD candidate (Media) at UNSW Australia

      In reply to Wil B

      "I don't know about you, but the risk of me getting killed in a car bomb, or tailed by a private detective, is rather low. So what and who cares if I have predictable movements. Why is this a bad thing???"

      This might seem fine for you, personally, but different demographics experience varying levels of threat from these and other dangers.

      If most of us feel we have noothing to fear and we increase the social reliance on such sytems, we marginalise groups who don't feel participation is safe for whatever reason.

      The article asks "So what should you do if you want to make your own daily patterns harder to predict?" The relevant part here is the 'if'. We're being shown some safety tips _if_ we are worried about our data. And the reality is that a significant number of people are, even if that percentage is not high.

      report
    2. Wil B

      B.Sc, GDipAppSci, MEnvSc, Environmental Planner

      In reply to Andrew McNicol

      yes, in Australia the booger wooger monster wonsters are out to get us all !1!!

      The only way to avoid getting tracked is to go and buy a roll of alfoil and carefully craft a hat to place on your head. Here's how: http://zapatopi.net/afdb/

      report
  2. Craig Morton

    Director at Moving Data - Telecommunications Stream Analytics

    Locational privacy is particularly challenging because much debate about privacy has traditionally been characterised in terms of who you are or what people are doing rather than where they are (and where they have been or where they are going).

    There are no specific provisions in Australian (and New Zealand) privacy legislation – location is
    not specifically identified as of concern. It is my understanding that globally only the UK regulation has considered location as an attribute of privacy…

    Read more
    1. Matt Duckham

      Associate Professor at University of Melbourne

      In reply to Craig Morton

      I agree with your points on the legal ambiguity of location data, Craig. I think there is particular ambiguity about the status of anonymized location or trajectory data (e.g., GPS tracking data about the movements of an unknown individual). Several studies have shown how easy it can be to infer a person's identity from this type of data (i.e., identity can be "reasonably ascertained"); yet such data is frequently not treated as "personal information." Second, I agree with your ideas of making more explicit the balance benefit versus privacy intrusion. Pizza delivery is a good example: it would seem entirely appropriate that informed individuals choose how much they value their location privacy versus having to travel themselves to pick up the pizza. I think a number of research studies and emerging technologies are attempting to enable individuals to enter into this sort of contract, as a complement to legal safeguards.

      report
  3. Les McNamara

    Researcher

    This article seems to be saying that most peoples' "workaday" movements are highly predictable and that if we go to more places and vary the order in which we visit places our movements will be less predictable, but this will be inconvenient. I'm struggling to find something remotely surprising about this result.

    report
    1. Matt Duckham

      Associate Professor at University of Melbourne

      In reply to Les McNamara

      For me, the surprising result of this study was quite how predictable workaday movements are. A remarkably small range of patterns (17 motifs, classified using just four rules) can describe the vast majority (90%) of individual's daily movements. And knowing one day's pattern for an individual is a very strong predictor of that person's future patterns (in the paper it is 10-30 times more likely than not that an individual will follow exactly the same pattern in the following day).

      report
    2. Les McNamara

      Researcher

      In reply to Matt Duckham

      Hi Matt. Without exaggeration, I can say that most people I know sleep, eat, commute, park and do chores like drop and collect kids at school at the same time, in the same way and at the same place each and every day. Many even eat the same breakfast, sit in the same carriage or seat, smoke the same number of cigarettes, eat at the same lunch spot and shop at the same supermarket at predictable times day in, day out.

      report