UNICEF-Logo (1).jpg




Estimate poverty rates across borders using social media data

Imagine what we might be able to achieve if we had up to date on poverty. What might a global solution look like that enables policy makers, businesses, and NGOs to understand where to direct resources to those in need? 



Challenge Briefing for CSE6242 / CX4242 Data & Visual Analytics students at Georgia Tech College of Computing 


Welcome to the UNICEF Data Innovation Challenge!


This opportunity has been curated by the UNICEF Innovation Fund specifically for students in Data & Visual Analytics at Georgia Tech. By successfully completing this project, you will be issued with a Certificate of Professional Development by UNICEF, certifying your skills in applying data to solve international development challenges. Students who submit extraordinary solutions may receive publicity as rising stars in the data field, and may also have the opportunity to work alongside UNICEF to bring your solutions to life. 


Innovation is vital to improving the state of the world's children. The speed at which global problems -- from disease outbreaks, to the global refugee crisis, to millions of out-of-school children -- disrupt the lives of children around the world is only getting faster. UNICEF innovates in order to stay agile and find solutions to the evolving challenges affecting all children."

- UNICEF Innovation team


Accurate and timely data can help us to understand and alleviate poverty, yet traditional methods of poverty assessment create major gaps in information. 


Accurate and timely estimates of population demographics are vital in order to understand social and economic inequalities. Such data shape decisions about which policies are implemented as well as where governments and humanitarian organizations choose to allocate scarce resources, especially with respect to the Sustainable Development Goals. Representative national surveys, the main method of collecting such demographic information, however, are both time-consuming and costly. As such, national surveys of socio-economic indicators are only performed every 5 to 10 years, and in some countries less frequently, up to once every 20 to 30 years.


In middle and high income countries, novel sources of passively collected data-sources have opened up new approaches to quantify and model vulnerabilities, inequalities, and large-scale population demographics. Previous work has used call detail records and data collected from social media to model poverty, dynamically map populations and understand unemployment. Although a majority of work has focused on modeling socio-economic indicators within individual nations using using varying types of data, little work has focused on understanding whether its possible to build socio-economic estimators for multiple countries from the same data-source or whether features are generalizable across models and nations. 


This challenge poses a powerful question with potentially high impact consequences: How might we come up with more accurate and timely data reporting mechanisms to monitor and alleviate poverty? 



Accurate and timely data can help us to understand and alleviate poverty, yet traditional methods of poverty assessment create major gaps in information. 

Our research indicates that anonymized and aggregated social media data from Twitter can be used to predict the Human Development Index (HDI). Ultimately, this can help us achieve the goal of bridging the gap between years where sub-national surveys are unavailable. The HDI is a composite index created to emphasize other development criteria than just economic growth alone; variables include life expectancy, education, and standard of living.


Our results show that the rich information captured by social media platforms such as Twitter enable us to build mathematical models of sub-national HDI using variables such as the adoption of the social media platform, the communication patterns of individuals, and patterns of travel. We show that this is the case

even for countries in which penetration of social media is low and both for developed and under-developed countries. 

This opportunity could allow us to look at the temporal evolution of socio-economic estimates, and to operationalize such models, enabling policy-makers and humanitarian organizations to base their decisions on up-to-date data.


Before taking on the assignment, read the full overview of the challenge as detailed by UNICEF data scientist Vedran Sekara: Predicting Socio-Economic Indicators for Multiple Countries Using Social Media Data. 


Create a visual interface, where the user can explore these insights: explore the data (eventually connect to an API), the different indicators that we calculate out of the data (i.e. mobility, activity), as well as the insights (poverty out of HDI (human development index).

Your mission is to develop the future interface for visualizing poverty in real time. How might we get from where we are now, with data collecting efforts coordinated by international organizations over multiple years, to the solution made possible by the information age: Understanding poverty as it exists today and as it is changing from day to day. Starting with social media, how might we assess and visualize the data available to us to understand poverty to the best of our current ability in the information age?​


You will use the following datasets while taking on the challenge: 


  • Real HDI (human development index), which measures the quality of life.

  • Estimated HDI, estimate using social media data (from Twitter).

  • Temporal patterns, i.e. when do people use social media (i.e. tweet).

  • Network data from mobility patterns, i.e. what is the flow of individuals between sub-regions of the country.

  • Predictability data (quantified using entropy), i.e. how predictable are the patterns emanating from each sub-region.”


By participating in this challenge, you will be taking a direct role in making an impact towards United Nations Sustainable Development Goal (SDG) 1: No Poverty, which aims to end poverty in all forms everywhere by 2030.  


By building the first visual interface to report poverty, you can contribute to a world where decision makers have the information they need to understand and address poverty in real time. Ultimately, this can contribute to faster rates of poverty alleviation, as well as additional support going to those whose socioeconomic conditions are not being adequately met by the international community.


Learn more about SDG 1: No Poverty 


Once you sign up, you will be verified by the ImpactEd team and provided with access to the online portal where you can browse the full suite of resources, challenge tips, and timelines.


Signing up does not require you to enroll in the challenge. This form simply allows us to contact you with more details about how to get started with the challenge, which will launch on September 15. 

Thanks! Message sent.