Call Detail Records (CDRs) can be used to produce a broad range of different indicators with useful applications in the humanitarian and development sectors.
From understanding the change in the geographic distribution of a population (dynamic population mapping) to determining between which areas, and when, people do travel (flows between locations), CDR indicators can provide an up-to-date and dynamic perspective on subscriber mobility within a country while preserving the individual privacy of subscribers.
These CDR indicators may be corrected to address biases in the data. Mobile device ownership and usage may differ based on factors such as age, gender or socio-economic status, which may affect the representativeness of the CDR data and any derived indicators. By supplementing CDR data with census or survey data, we can better understand the variation in mobile device ownership and usage, and adjust the indicators to address any biases.
Dynamic population mapping indicators
Indicators on the changes in the geographic distribution of a population may describe short-term (hourly, daily) changes in the number of people visiting locations, referred to as ‘subscriber presence’, or long-term (weekly, monthly, seasonal) changes in where people reside, described as ‘home locations’.
Short-term changes in the geographic distribution of a population can also be used to quantify the “crowdedness” of locations, how many people visit a location and how often large numbers of people visit a location, and “population mixing”, the number of people visiting the same locations in the same time period.
However, while CDRs can provide estimates of re-distribution of the population over time, it is important to note that it cannot provide information on how the population is growing or shrinking overall, due to births, deaths and international migrations.
Human mobility indicators
We can describe different aspects of human mobility through a broad range of CDR indicators. Short-term mobility (hourly, daily) may be described by the number of journeys between pairs of locations (trips) or the number of journeys within, into or out from different regions (intra- and inter-regional travel). We can also produce indicators to describe the “connectivity” between different regions to identify clusters of connected areas, areas which are isolated and main travel routes between regions.
We can also describe longer-term (weekly, monthly, seasonal) aspects of human mobility, in particular the number of people who changed their home location (home relocations). This allows us to produce indicators describing the number of people relocating between regions and which regions are gaining or losing residents.
Furthermore, when CDR data is combined with other types of mobile operator data or additional features, further indicators can be calculated. For example, if the CDR data provided for analysis also contain corresponding information for the other party to a call or SMS message, we can derive social network indicators, such as the social connectivity between different areas. Furthermore, the geographic distributions of social networks are an important predictor of where people are displaced to during a disaster, meaning this information can contribute to disaster preparedness.
In order to support the interpretation of the indicators, we can also produce diagnostic indicators. These describe changes in mobile network activity such as the number of network events, the numbers of subscribers in each area regularly using their mobile devices and the number of new subscribers. While they do not directly describe mobility, these indicators allow us to control for changes in subscriber behaviour which might otherwise be mistaken for changes in mobility.