Smith-Clarke, Christopher; (2021) Estimating poverty maps from aggregated mobile communication networks. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Smith-Clarke_10122237_Thesis.pdf Download (8MB) | Preview

Abstract

Governments and other organisations often rely on data collected by household surveys and censuses to provide estimates of household poverty and identify areas in most need of regeneration and development investment. However, due to the high cost associated with manual data collection and processing, many developing countries conduct such surveys very infrequently, if at all, and only at a coarse level of spatial granularity. Consequently, it becomes difficult for governments and NGOs to determine where and when to intervene. This thesis addresses this problem by examining the feasibility of deriving up to date and high resolution proxy measurements of poverty from an alternative source of data, namely, Call Detail Records (CDRs), which can be used by organisations to help in decision making. Specifically, we contribute the following: 1. A detailed spatial analysis of economic wealth in two sub-Saharan countries, Senegal and Cote d’Ivoire from which we derive two baseline poverty esti- ˆ mators grounded on concrete usage scenarios. 2. We establish a link between communication patterns and wealth through a simulation-based analysis of information diffusion. We further examine the influence of contextual factors, including data quality issues and economic volatility, on the strength of this relationship. 3. An approach to building wealth prediction models based on features of aggregated CDRs. Features include static and simulation based measures of information access, activity based metrics and econometric inspired metrics. We further perform a comparative analysis of the results of several models in relation to the baseline predictors. We conclude that it is possible to produce proxy poverty or wealth indicators from aggregated CDRs that provide a good level of accuracy, particularly where geographical coverage of the mobile phone network is sufficient. The final outcome of this thesis is a method for developing aggregated CDR-based poverty or wealth models that can be readily implemented anywhere in which there is a need for more up to date and/or finer resolution poverty estimates.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item