Machine learning for population growth modeling through annual CO2 data

Abstract

This article presents a general framework to model the population growth of Colombia through machine learning, which allows incorporating variables such as CO2 emissions. The data are obtained through the DANE Geoportal and the World Bank portal database. In recent years, several cities in Colombia have experienced an accelerated population and urban growth as a result of an overwhelming increase in the immigrant population, a situation that has been affecting the urban order of the city, mainly due to population confinement, increased environmental pollution, unhealthiness, poverty among other factors.The population growth prediction fully shows the effects of the fertility rate, mortality, and migration of people from one territory to another, essential factors in determining population and health system conditions. Accurate information can be obtained from population census measurements, a strategy with a high cost; However, the capabilities of machine learning are growing, being also implemented to determine predictions of systems and behaviors in demographic, environmental, and socioeconomic issues, which contributes to the economic and energy-saving that implies the gathering and processing of demographic information for the country.In this work, to predict population growth, the concept of machine learning was applied, carrying out a data analysis using an Artificial Neural Network (ANN). The results obtained provide a predictive model with the potential to provide regional and governmental entities with suggestions, strategies, and alternatives for studying the population and making decisions for urban redevelopment.