Abstract
This study is about the prediction of life expectancy in different continents and countries and the analysis of influencing factors to determine the factors that lead to the increase or decrease of life expectancy. There are 24 variables from 2000 to 2015 for nearly two hundred countries across six continents. These variables include mortality, economic factors, social factors, and immunological factors. The analysis of lasso parameter table shows that GDP, adult mortality rate, and percentage expenditure generally have the greatest impact on life expectancy in six continents. The life expectancy in North America is least affected by factors, while the life expectancy in Asia is most affected by factors. Analysis of random forest model shows that feature importance of six continents is different. In addition to adult mortality and income composition of resources, HIV/AIDS in Africa (0.4) and thinness 5–9 years and thinness 1–19 years in Europe are the most significant. In the XGboost model, the lowest SHAP values were found to be Haiti in North America and Sierra Leone, Botswana and Eritrea in Africa, with life expectancy below 42, HIV/AIDS, Adult mortality and income composition of resources have the greatest influence on their life expectancy. In the samples with the highest SHAP value, it is found that the life expectancy of Canada in North America and Ireland and Germany in Europe is above 80, which are also affected by the same factor and thinness. In the interaction analysis, it can be seen that HIV/AIDS is negatively correlated with income composition of resources and adult mortality. The higher the number of years schooling is, the higher the SHAP value is. The higher the income composition of resources, the lower the adult mortality.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call