Machine Learning-Based Life Expectancy Prediction in Developed and Developing Regions
Battina Reshma
Life expectancy is a core demographic and public health indicator reflecting the overall quality of healthcare, living conditions, nutrition, socioeconomic progress, and disease control in a country. However, manually analyzing how dozens of health, social, and economic indicators influence lifespan is complex and time-consuming. This paper presents an interactive, web-based intelligence application developed using the Django framework, SQLite database, and machine learning regression algorithms to forecast life expectancy in developed and developing regions. The system utilizes real-world health and development indicators from the World Health Organization (WHO) and United Nations. We implement and evaluate three regression models: Linear Regression, Random Forest Regressor, and XGBoost Regressor. Preprocessing steps handle missing data, normalize features, and split records into training and test sets. XGBoost achieves superior predictive performance due to its robust gradient-boosting trees, yielding the lowest Mean Absolute Error (MAE) and highest R-squared ( R 2 R 2 ) value. The system is integrated into a Django web platform, allowing users to view data tables, train models, compare performance metrics via interactive visualizations (correlation matrices, scatter plots, violin plots, and radar charts), and predict life expectancy through user-friendly forms.