Skip to content

Introduction

Motivation

Today, just over half of all people worldwide use the Internet. Yet in our fast-paced digital world, being offline excludes people from a host of opportunities. The Internet has grown beyond the remit of mere communication service and now grants access to global knowledge, social empowerment and new economic activities. In a word, the Internet fuels social and economic development and thus has been moving higher on national policy agendas around the world, all the more in the aftermath of the COVID-19 pandemic. Moreover, many countries lack data on Internet use in the society and can therefore not tackle the challenges most efficiently. Giga, the joint initiative by the International Telecommunication Union (ITU) and UNICEF, has set the goal to connect every school in the world to the Internet by 2030. Giga aims at giving children and youth the opportunity to get online and tap the potential of the Internet to their benefit. Connected, digitally savvy and enlightened youth can lead economies and societies towards a more productive, fulfilling, equal and sustainable future. What’s more, as schools often serve as community hubs, the worth of connectivity can be extended to larger groups of the population.

Arguably, connecting schools in no specific order does not allow for allocating resources where they are most needed and connecting communities that are most in need in a timely fashion. Giving priority to connecting schools that are closer to nodes or pro-actively apply for a connection may not serve national policy priorities best and could potentially even deepen gaps between urban and rural areas or across demographic groups, since the offline population is notoriously hard to target. This project therefore aims to provide national decision makers, multi-lateral agencies and development partners with a sound rationale for prioritization of schools to be connected, building on a large range of complementary datasets, advanced modelling and machine learning to identify the largest offline communities and craft connectivity roadmaps based on the most recent, relevant and accurate evidence available.

Project objectives

This pioneering project aims at developing mechanisms for predicting the world's offline population, using school locations as the starting point. A multi-disciplinary team of fellows hosted by the University of Warwick jointly with experts from ITU and UNICEF have examined three case studies, namely Brazil, Thailand and the Philippines. Based on Big Data and machine learning techniques, the team has created a toolbox that can be used by researchers and institutions to investigate the state of connectivity within a country and enable near real-time national and regional benchmarking. The project outcome can be readily reproduced for other countries and the individual models can be enhanced for specific purposes. The evidence generated by the models can support digital policy planning and implementation at the national level and inform the work of all national stakeholders and development partners involved in connecting the unconnected at the sub-national, national and regional levels. Concretely, the ambition of this project has been three-fold.

  • First, to estimate the share of households/individuals with Internet connectivity around a school. Internet connectivity is defined here as any ability to get online, either through a fixed or mobile broadband connection. This also means that individuals must own or have access to broadband-enabled devices. Based on population data for the area around each school, the offline population across communities is estimated. A rank of priority can then be given to communities and a recommended order in which schools can be connected in view of optimal resource allocation and fast-tracking the achievement of social and economic policy objectives. In addition, quick wins can be identified through further modelling and analysis, also taking into account data on computer availability and electrification in schools.

  • Secondly, based on the best estimates of community level connectivity, to aggregate Internet connectivity data up to the country level. While such aggregate figures might already exist for some countries, this bottom-up approach will be helpful in countries where there are currently no census or surveys or other established statistics at hand. Furthermore, an elaborate prediction has the potential to even be more accurate or more complete than a national survey. The model establishes a baseline metric for a country and builds in-depth understanding of countries’ level of Internet connectivity. The geospatial analysis integrated in the modelling also reveals regional differences and local peculiarities.

  • Ultimately, this work contributes to the field of offline population research and provides what is, to our knowledge, the first bespoke predictive model in this field. We have gained insights on the features and models that are most efficient in predicting Internet connectivity, so that national decision makers can fine-tune their strategies and own analysis while researchers and organizations can continue building on these initial findings. All processes, analysis and findings are duly documented and available as open data [1]. The collection of scripts and documentation can be cloned via the GitHub repository (https://github.com/DSSGxUK/itu).

[1] National datasets are not part of the publicly available data package. National decision makers can nevertheless choose to also release their datasets and thus contribute to the replicability of the model and the global research community’s enhanced understanding of the topic.