META-ANALYSIS OF USAGE OF MACHINE LEARNING TO IDENTIFY AND ASSESS POVERTY

Abstract

Poverty is dened as a person’s state of survival without the adequate nancial resources for a minimum standard of living. Poverty is assesed through multidimensional factors like access to clean drinking water, access to electricity, access to quality education, child mortality rates, nutrition, and so on. On 1st January 2016, the United Nations Economic and Social Affairs department published a list of 17 Sustainable Development Goals (SDGs), and “No Poverty” was the rst goal on the list and the purpose was to eradicate all forms of poverty from all corners of the globe by the year 2030. The rst step in the goal is to identify poverty in all its forms and not just the income level. The initiative of identifying poverty is a humongous task in itself and various researchers, academicians, statisticians and computer scientists had proposed several methods of identifying poverty in all it’s forms. This paper unies and consolidates several machine learning techniques proposed previously to theoretically formulate a new, robust methodology to identify, validate and assess poverty which would be the rst step towards sustaibale development.