Identifying population movements using anonymised telephone data

This article was published on 12 December 2018.

Major metropolitan areas in the Netherlands routinely face the problem of sudden increases in crowding in town and city centres. That impacts areas such as mobility, safety, emergency response, and tourism. To manage this effectively, policymakers need more detailed information about the number of individuals present at any given location. They also want to access this information more quickly. New research from Statistics Netherlands’ (CBS) Center for Big Data Statistics (CBDS) offers a potential solution. Previous research by CBS showed that mobile telephone data offers a promising solution for determining how many individuals are present in a given location during the day. Uniquely, this source can be used to estimate the number of people present at any given moment throughout the year, even in small areas. It provides detailed, readily available information that researchers can use to monitor the dynamics of a population from hour to hour.

Large datasets

For the purposes of this article, CBS used anonymised mast data from T-Mobile. The datasets in question are very large indeed. This is because mobile phones connect to different masts many times each hour. They can do so up to 600 times per day, and, at any given time of day, millions of phones are operating simultaneously on T-Mobile’s network. Thus, in just a single day, data centres will record the details of billions of mast connections. This data is then processed into statistical information.

Every day, the total distance travelled by people in the Netherlands averages 532 million kilometres. Each individual spends 58 minutes travelling, on average. Based on data registers, such as the Municipal Personal Records Database, it is possible to create statistical analyses about people’s place of residence. However, little is known about people’s movements during daytime hours. The same is true of the dynamics of the population throughout the day, or in a given month. However, details such as these are an important source of quantitative information in the areas of mobility, safety, emergency response and tourism. In these areas, there is a great need for readily available, more detailed data about where people are located.

Method

CBS developed this method in close cooperation with T-Mobile. This cooperation ended in december 2019. To arrive at the counts, we have to complete a series of steps: First the data are made completely anonymous, right at the source (T-Mobile). A detailed description of the method used will be made available soon.

When processing this data into statistics, privacy is guaranteed. This is because we only work with anonymised, aggregated data. Such data cannot be traced back to the level of individuals; it is always in the form of anonymised datasets. A range of measures have been taken to exclude any possibility that the data could be traced back to specific individuals. For instance, all of the data is retained at the T-Mobile data centres, and the focus is on series of events (activity) within the network.

Findings

Details of visitor patterns per local authority are shown on a dashboard. When users click a given location, a pattern is projected for the local authority in question, showing the number of people entering that local authority area, and the numbers leaving it. The pattern shows the levels of crowding in a local authority area, over time. In this way, it could be considered to represent the “pulse” of that local authority.

Interactive map display

In some cases, the level of crowding in a local authority area is influenced by the weather. For this reason, the visualisation displays details of the weather conditions at the top left. At the top right, in the legend, it is possible to select a given category, for example people entering the local authority area or, indeed, those leaving it. It is also possible to launch an animation (at the bottom left of the map) showing the rise and fall of the population over time, in the local authority area concerned.

Your feedback is welcome