Uma análise abrangente do índice de desenvolvimento humano utilizando técnicas de machine learning

Abstract

The Human Development Index (HDI) is a comprehensive indicator aimed at measuring social well-being in various countries and territories. It was conceived as an alternative to measures that rely solely on monetary aspects. Designed to be a simple measure, the HDI utilizes only three basic dimensions to assess the quality of life of the population: health, education, and standard of living. In this context, this study aims to advance the understanding of which factors, despite not being part of the index, are associated with the HDI and can contribute to its improvement. Specifically, the aim is to understand the factors influencing the Human Development Index (HDI), using data modeling collected from official sources in 191 countries. As explanatory variables, factors such as CO2 emissions, female participation in the workforce, Gini index, prevalence of mental disorders, indicators of religious, moral, and sexual freedom, and renewable energy consumption, among others, were considered. To investigate the relationship between the HDI and explanatory variables, random forest models, a Machine Learning technique, were employed, using both regression models to predict the HDI value and classification models to categorize it. The obtained models demonstrated considerable predictive capacity. The regression model presented a coefficient of determination (R2) of 0,932 and a root mean square error (RMSE) value of 0,042. Meanwhile, the classification model achieved an accuracy of 73%.