New geographic data science methodologies and applications

We are developing new geographic data science methodology that aim to extend the techniques of data science with a more critical and nuanced understanding of geography and spatial processes.

Estimating displacement in real time

With nearly 110 million people forcibly displaced worldwide, understanding where people move - and how fast - is critical for effective humanitarian response. Yet traditional data sources are too slow, too costly, and too infrequent to support real-time decision-making.

We developed a novel framework that harnesses anonymised mobile phone location data from 25 million devices to produce population-level estimates of internal displacement. Applied to Ukraine during the early stages of the 2022 Russian invasion, our method delivers validated, daily estimates at fine geographic resolution - enabling humanitarian and disaster management organisations to prioritise resources where they are needed most.

The framework is designed to be portable: given comparable mobile data, it can be deployed to monitor displacement driven by conflict, climate hazards, or epidemics anywhere in the world.

Publication

Iradukunda, R., Rowe, F., & Pietrostefani, E. (2025). Estimating internal displacement in Ukraine from high-frequency GPS mobile phone data. Humanities and Social Sciences Communications, 12(1), 1863.

Assessing and making digital trace data more representative

Traditional sources of population data, such as censuses and surveys, are costly, infrequent and often unavailable in crisis-affected regions.

Mobile phone application data (MPD) offer near-real-time, high-resolution insights into population distribution, but their utility is undermined by unequal access to digital technologies, creating biases that threaten representativeness. Despite recognition of these issues, no standard framework exists to address such biases, limiting the reliability of MPD for research and policy.

We develop and implement a systematic, replicable framework to quantify and explain population coverage bias in aggregated mobile phone application data without requiring individual-level attributes. The approach combines an indicator of population coverage bias with explainable machine learning to identify contextual drivers of spatial variation in bias.

Using four datasets for the United Kingdom benchmarked against the 2021 census, we show that MPD achieve higher population coverage than national surveys, but biases persist across sources and subnational areas. Population coverage bias is strongly associated with demographic, socioeconomic and geographic features, often in complex nonlinear ways. Contrary to common assumptions, multi-application datasets do not necessarily reduce bias compared to single-app sources.

Our findings establish a foundation for bias assessment standards in MPD, offering practical tools for researchers, statistical agencies and policymakers.

Publication

Cabrera, C. and Rowe, F. (2025). A systematic machine learning approach to measure and assess biases in mobile phone population data. arXiv preprint arXiv:2509.02603.