Forecasting GDP Per Capita Using Machine Learning Algorithms

Traditional GDP per capita calculations were based on simplistic models, often relying solely on population figures as the primary determinant. However, as our understanding of economics and data analysis has evolved, we have come to recognize that a multitude of socio-economic and demographic factors significantly influence a nation's economic well-being. This limitation leads to imprecise predictions and hampers the ability to make data-driven policy decisions and strategically plan economic development. In response to this challenge, we propose a pioneering approach that harnesses the capabilities of Machine Learning (ML) to model these intricate and non-linear relationships, significant improving the accuracy of GDP per capita forecasting. Within our framework, we have taken into account factors such as population, area, population density, coastline, net migration, literacy, phones, infant mortality, arable land, crop land, other land, birth rate, death rate, region, and climate to compute the GDP per capita of the country. Our study involves a comprehensive comparative analysis of two robust ML models, Random Forest and Neural Network regression. These models excel in handling complex data relationships, uncovering nuanced patterns, and effectively mitigating common issues like overfitting. By systematically assessing the performance of these models, we utilize critical metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), and R-squared (R²) score. The insights gained from this comparative analysis are invaluable for policymakers, economists, and researchers, as they facilitate more informed and evidence-based policy decisions, optimize resource allocation, and enhance economic planning.