Methods: Mobility indicators
Page under development: It will be updated with method details for each indicator.
Using a combination of the CDR aggregates described here, we can generate a range of mobility indicators which are useful for planning and evaluating interventions related to mobility in the context of the COVID-19 response.
Analysis, including data cleaning, aggregation of data across time periods, or rescaling to account for biases within the data, is required to produce these indicators.
Most indicators are formed using the following standard operations:
Computing a baseline ( for comparison or for computing a % change)
Computing a percentage change compared to baseline (often but not always)
Rescaling the data to account for changes in phone usage (active subscribers and number of events)
To do this all CDR aggregates need to be computed for a period of at least four weeks before interventions were introduced, or before the initial COVID-19 cases were announced in the country. This is to obtain indicators about routine baseline mobility, against which the post-intervention (or post-news) indicators can be compared. CDR aggregates should then be produced continuously throughout the epidemic, and output with a frequency that meets the needs of the data consumers.
To interpret and visualise the indicators the boundary datasets used to produce the spatial aggregation needs to be provided along with the CDR aggregates. We also provides guidelines to MNOs to aggregate their towers into clusters or by administrative regions here.