Introduction
Poverty is a global problem that affects hundreds of millions of people worldwide. Many people cannot afford the common goods/services we need to survive: food, water, healthcare, etc. According to the World Bank, approximately 689 million people live in extreme poverty meaning that they live on less than $1.90 or less a day. These are just the people that have been accounted for, but there could be millions more. Solving the global poverty crisis has been one of the United Nation’s (UN) long-term goals for decades; however, they have used outdated technology and analysis to tackle the problem. To truly solve the global poverty crisis, we need to truly gauge the severity of the problem efficiently. Gathering viable data about poverty has been an issue for many researchers, as the regions that have been struck by poverty are usually located in territories of war or conflict. In this literary review, we will assess how poverty is mapped and more importantly how those results can be analyzed to help solve the poverty crisis in Eastern Africa.
Why Eastern Africa?
I chose to assess poverty in Eastern Africa, because this is an issue that is very important to me. I have many friends of African descent, whose families still reside in Eastern Africa (Somalia, South Sudan, Kenya, Uganda). I have seen pictures and videos of the quality of life there, and it is absolutely disgraceful that people are being allowed to live like that. Many people take for granted all the resources that they have, but do not realize that others in regions such as East Africa have very little if not any. I believe greatly in human development; As technology and resources increase, so should quality of life and the freedoms that many people enjoy. Amartya Sen, a famous Indian economist, and philosopher who is famous for his book: “Development of Freedom”, is an expert in the philosophy behind human development. His definition of development is the process of expanding the freedoms that people enjoy. With the process of expanding freedoms, also comes the removing of sources of unfreedoms: tyranny, poor economic opportunity, poverty, etc. I agree greatly with his definition of development, and I feel that with poverty comes many of the unfreedoms he talks about. Eastern Africa is one of the regions that struggle most with development. Many of their people do not even have basic freedoms (facilities for education, health care, industrialization, social modernization) that Sen talks about. The story of Kader Mia is one that reminds me of how severe the problem of poverty is in Eastern Africa. Kader Mia was stabbed because he was trying to get to work in order to support his family. He and many others are trying to escape their unfreedom of poverty, but it ends up costing their lives and/or their livelihood. There is no question that poverty is a major problem, specifically in Eastern Africa, but our newfound knowledge of data science/analytics may help us solve this problem.
The Research/Data Science Methods
Jessica Steele is a data science researcher who really helped me understand how something as simple as a phone can help map out poverty. Mobile phones are capable of creating high-resolution poverty maps, and researchers found out that if mobile data were to be combined with geospatial data from satellites, then poverty predictions could be produced. The data that is being collected from the phones can range from: apps to data usage to even the number of texts sent/received. All these individual sets of data, when combined can be used to paint a picture not yet conceived. Steele delves deeper into this process by providing technical terms of RS (Remote Sensing data) and CDR (Call data record) that helps show how a simple phone can be used to map out poverty. RS data or Remote sensing data collects physical/geographical properties of land, while CDR or call data record data collects information at the level of cell towers. This data can be used to assess asset consumption and other income-based measures of wellbeing. In Steel’s study we saw that the model that used both CDR and RS data performed exceptionally in urban areas with an r2 value of 0.78 and rural areas as well with an r2 value of 0.66. At the national level, the model performed even better with r2 value of 0.76. Though the performance of RS-only or CDR-only models are still very impressive, it is heavily dependent on context. Steele’s application of RS and CDR data were modeled in South Asia, but its potential to be used in regions such as Eastern Africa is still being explored. Dr. Steele is a great researcher to consult in respect to RS and CDR data, but I also find Utz Pape’s usage of satellite data and geospatial techniques to analyze poverty in inaccessible areas to be very interesting too. The regions that Pape was analyzing were South Sudan and Somalia, two countries in East Africa that have been overcome by great conflict and war that has caused immense poverty. These countries are highly data-deprived, so to fill this gap Pape inducted the use of the HFS (High Frequency South Sudan Survey) for South Sudan and the SHFS (Somalia high frequency survey) for Somalia. Both surveys were conducted via the same method, by using geospatial techniques and high-resolution imagery to model population distribution and build a probability-based sampling frame. All this geo-spatial data allowed for the creation of a “security assessment access map”, which allows researchers to analyze the conditions in inaccessible/unsafe areas. It was very interesting to see how satellite and geo-spatial data could be used to create poverty prediction models in such a safe and non-invasive way. The models that were created were even capable enough to employ special strategies to analyze the conditions of many nomadic populations that are prevalent within Eastern Africa.
From all of the articles that I have analyzed and read, there is one thing in common with all of them; It is the criticism and outdatedness of the census. Before reading these articles and being introduced to the realm of data science, I thought that the census was a crucial and powerful tool used by many nations to assess economic prosperity and overall well-being. However, now it just seems like an outdated and ineffective tool. Dr. Jessica Steele states “Census and household surveys are normally used as data sources to estimate rates of poverty. However, they are not regularly updated and in low-income countries, surveys can be patchy,” I agree completely with this statement, and I feel that the new data analytic methods (CDR, RS, Satellite-data, Geo-spatial data) are far more efficient and useful than the census. The information is consistently updated, while the Census takes place every 10 years.
Now we come to the data science methods and datasets used to create these detailed poverty maps. There are a various number of data science methods that go hand in hand when producing poverty maps. There is Remote Sensing (RS) data, CDR (Call data record) data, Satellite-data, Geo-spatial data, etc. These methods are used to analyze a various datasets, and each method has unique data sets which they analyze. CDR and Satellite- Data, in respect of the studies that I analyzed today assessed these data sets: call time, data usage, and number of texts sent/received. This information can be useful, in that machine learning can tell us how much a certain region is spending on phone usage or not spending on phone usage. This can be used to show gaps within a map, where there is not as much phone activity. RS data/Geo-spatial data/Satellite imagery are some of the data methods that are especially interesting. They gather data ranging from: conditions of roads/homes, population distribution, etc., and then this is created to predict poverty within the certain region that is being assessed. This is very interesting, in that a machine can learn and predict poverty simply by looking at day-time images and then displaying a map in a night-time setting that looks like a “heat map for poverty”. Other data science methods such as clustering and predictive modeling are also included in the various methods I discussed. All of these methods coupled together not only map out poverty, but depict the quality of life, economic prosperity, health of a community, and much more. Something that I do not like about the Census is that most of the time all it does is just state or map out information. It just shows that a certain region has crippling poverty but does nothing in response to that. Utility of Findings
Something that we need to be careful of as our techniques for data analysis advance, is to maintain vision on the goal: to solve the poverty crisis in Eastern Africa. Obviously being able to predict poverty and map out regions that were not able to be mapped out before is an amazing feat, but the next step is to use that information in an efficient way. With all this information gathered through various data science methods, proper aid and funding can be administered to regions in need efficiently and effectively. New businesses can start up and provide individuals with job opportunities that can help sustain them and their families. Though Africa is behind the rest of the world, the region is slowly starting to become industrialized. Companies like Jumia (a company that I currently invest in and am very fond of) is considered to be the ne2xt Amazon of Africa, and this can provide many Africans with new jobs. Firms in U.S and Europe openly invest in African companies like these because they know that it is a win-win situation for all countries. Job opportunities within Africa can grow, and the overall economy can benefit from this new source of e-commerce. Additionally, Governments can use the data produced by the poverty maps and predictive modeling in order transfer funds effectively and build schools, in order to increase the level of education. Poverty is a mix of many unfreedoms: lack of education, infrastructure, food, water, supplies, etc. Now that there Poverty can be eradicated, the proper infrastructure and funding must be laid out in order to remove these unfreedoms and expand new freedoms. After reading many of these articles, I have come to learn that Poverty is a combination of: Social, Economic, and Environmental features. Specifically, in East Africa, the environment is poor in that access to jobs in safe, clean water, and food security are all limited. Socially, many of the residents are poor and do not have anyone to rely on to escape poverty. Economically, there are no businesses or solid infrastructure that can allow for a large sum of job opportunities. In this assignment, we were assigned to choose a specific topic whether that be: food security, land use, urbanization, migration. However, I feel that all of these topics work hand in hand with each other, especially poverty. All these topics must be addressed and analyzed to expand the freedoms that people that enjoy and remove the ones that people do not.
Conclusion
Though, much of what I read was very comprehensive and detailed; there are still a few gaps in the literature that I think should be addressed. Like I said before, all of these articles dealt with the mapping and prediction of poverty, yet none of them addressed how that information was used. I would like to learn more about if these new techniques of poverty assessment and predictive modeling can be used all over the world, not only nations struck with extreme poverty. Data science/analytics has come a long way, yet still has much more untapped potential. We are coming into a new age of statistics and analysis, that will usher true development. First, we must start by removing the unfreedoms that exist such as the ones in East Africa. Then, we can focus on further developing our way of life. Amartya Sen has a vision, and it is one that the powers of Data science can achieve. I feel that the research question that all my analysis and reading has brought is: Can this data posed by the poverty maps and predictive models be utilized in nations where poverty is not as openly apparent when compared to the countries in Eastern Africa?
Citations
Castelan, C. R. (2019, July 9). Making a better poverty map. Retrieved March 22, 2021, from https://blogs.worldbank.org/opendata/making-better-poverty-map
Horton, M. (n.d.). Stanford scientists COMBINE satellite data, machine learning to map poverty. Retrieved March 22, 2021, from https://pangea.stanford.edu/news/stanford-scientists-combine-satellite-data-machine-learning-map-poverty
Kumar, A. (2020, July 06). How to understand global poverty from outer space. Retrieved March 22, 2021, from https://towardsdatascience.com/how-to-understand-global-poverty-from-outer-space-442e2a5c3666
Martinez, A., Jr. (2021, March 11). Using machine learning on satellite images to map poverty. Retrieved March 22, 2021, from https://development.asia/insight/using-machine-learning-satellite-images-map-poverty
“Pape, Utz; Parisotto, Luca. 2019. Estimating Poverty in a Fragile Context : The High Frequency Survey in South Sudan. Policy Research Working Paper;No. 8722. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/31190
“Pape, Utz; Wollburg, Philip. 2019. Estimation of Poverty in Somalia Using Innovative Methodologies. Policy Research Working Paper;No. 8735. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/31267
Steele, J. (2017, February 19). Mobile phones can create high-resolution poverty map. Retrieved February 23, 2021, from https://www.indiatoday.in/technology/news/story/mobile-phones-can-create-high-reolution-poverty-map-959791-2017-02-09
Steele JE et al. 2017 Mapping poverty using mobile phone and satellite data. J. R. Soc. Interface 14: 20160690. http://dx.doi.org/10.1098/rsif.2016.0690