Determinants of Global Life Expectancy: A Machine Learning Analysis Using Random Forest Models

In this paper, I adopt a data-driven approach to analyze global life expectancy. Using World Bank data from 174 countries covering the period 2001–2019, I model economic, social, and health factors that may predict a nation’s life expectancy. I first performed data preprocessing, which included median imputation for missing values, and then tested several regression models, as well as Random Forest, Boosting, and Decision Tree models. The Random Forest Regressor exhibited the best predictive performance and stability, as it can capture both non-linear and multi-level relationships between income and region. Feature importance analysis showed that undernourishment, communicable diseases, and healthcare expenditure had the most significant impacts on life expectancy, while education expenditure and injury had the least. These results support the view that income alone cannot guarantee population health; instead, investments in health infrastructure and access to basic resources are key to extending overall life expectancy.