From raw data to insights

Several key stages are required to convert raw CDR data into actionable insights. These are categorised as follows:

Raw Call Detail Records (CDR data)

Recorded by Mobile Network Operators

CDR data are owned by MNOs and are produced for billing purposes. The data stay within MNOs and is not accessed by other parties.

Raw CDRs are generated each time a mobile phone subscriber makes or receives a call/text or uses mobile data. Each record includes an anonymous identifier of the subscriber, a timestamp, and the cell tower that the transaction was routed through.

CDRs can be useful mobility resources as they contain a record of each subscriber’s approximate location (the location of the cell tower) each time they use their phone.

Raw CDR data are necessary to produce CDR aggregates.

Production of CDR aggregates

Produced by MNOs and Regulators using Flowminder's code on GitHub

CDR aggregates are produced by processing the CDR data of many individual subscribers into an output that characterises the behaviour of the entire group of subscribers. These aggregates can be produced by MNOs and regulators using Flowminder’s code, which is available on GitHub.

Examples of commonly-produced aggregates are the count of subscribers actively using their phone in a given region within a specified hour, or the count of subscribers travelling from one region to another on a particular day.

In order to ensure statistical validity, and ensure that information about any single subscriber is not inadvertently disclosed, Flowminder's code will only produce aggregates for groups of at least 15 subscribers.

Production of mobility indicators

Produced from CDR aggregates by analysts and modellers

Data analysts can extract information from CDR aggregates to produce descriptive statistics of mobility.

Flowminder analysts propose here a range of indicators that capture mobility characteristics that are relevant to the COVID-19 context.

Indicators can be interpreted by those who have knowledge of the local context. For example, a 30% decrease in daily trips to the capital city post-lockdown could be because the workers living outside the city have stopped going in.

Turning mobility indicators into actionable insights for decision making

Produced by skakeholders with contextual knowledge

Mobility indicators and measures of changes in population mobility are valuable for decision-makers to develop and implement the necessary interventions to protect their populations against the virus. It is also vital to know when these restrictions can be relaxed.

The interpretation of mobility indicators, incorporating local knowledge and additional data sources (such as locations of schools or shopping districts) can provide governments or health actors with a wide range of insights. For example, it is possible to quantify the effect of interventions on different types of movements (e.g. commuting, attending church, long distance travel), and to assess whether interventions result in unintended effects (e.g. large scale exodus from major cities). This then enables the simulation and planning of the implementation and relaxation of interventions, in stages.

Analysts can interpret the indicators to produce insights for decision makers. These can take the form of graphs, maps, statistics or reports, which can be used by governments, journalists and other scientists to benefit the response against COVID-19 at a national or global scale.

See more about applications and insights here.