Life Expectancy Regression Analysis

This paper explores the relationship between life expectancy and a range of factors, including vaccination rates, health status, and socio-economic conditions, through the application of a multiple linear regression model. The study encompasses initial data visualization and model assumption checking, addressing heteroscedasticity through techniques such as removing influential points, Box-Cox transformation of the re-sponse variable, Goldfeld-Quandt test, and Weighted Least Squares regression. Two optimal models are derived: one with the best explanatory capacity and another with the best predictive accuracy. In the discussion section, we suggest refining the model using decision trees, stratified by the development status of countries (developed ver-sus developing). This involves fitting distinct regression models for each group to identify more accurate predictive models. Based on these findings, we provide rec-ommendations for improving population life expectancy.