Better coverage, better outcomes? Mapping mobile network data to official statistics using satellite imagery and radio propagation modelling

Abstract

Mobile sensing data has become a popular data source for geo-spatial analysis, however, mapping it accurately to other sources of information such as statistical data remains a challenge. Popular mapping approaches such as point allocation or voronoi tessellation provide only crude approximations of the mobile network coverage as they do not consider holes, overlaps and within-cell heterogeneity. More elaborate mapping schemes often require additional proprietary data operators are highly reluctant to share. In this paper, I use human settlement information extracted from publicly available satellite imagery in combination with stochastic radio propagation modelling techniques to account for that. I show in a simulation study and a real-world application on unemployment estimates in Senegal that better coverage approximations do not necessarily lead to better outcome predictions.

Intercensal Updating Using Structure-Preserving Methods and Satellite Imagery

“…We aggregate the pixel values via their centroids to the respective administrative areas in Senegal. For discussions on the impact of allocation uncertainty across different area systems we refer to Koebe (2020) and Groß et al (2020). While WorldPop already provides population estimates for Senegal in their open data repository, we opt to slightly modify their approach to align it with our case study.…”

Leveraging Mobile Phone Data for Migration Flows

“…Researchers, for example, may use such data to support policymakers, organizations and authorities in taking data-driven decisions related to migration [25]. In the past, mobile network data have been used to solve a variety of societal challenges [23], including modeling migration, assessing the migrants' well-being (e.g., [26,27,28,5,29]) and estimating population displacements (e.g., [8,30,31,32]). This Section describes how mobile phone data have been used to monitor different types of migration.…”

Releasing survey microdata with exact cluster locations and additional privacy safeguards

“…This augmentation is usually done via geographic matching, i.e. combining area-level averages [14]. Since the number of matched areas corresponds to the sample size for subsequent supervised learning tasks, finding the smallest common geographical denominator is essential to avoid running into small sample problems.…”