Prediction of life expectancy for Asian population using machine learning ALGORITHMS / Nurul Shahira Pisal, Shuzlina Abdul-Rahman, Mastura Hanafiah and Saidatul Izyanie Kamarudin

Predicting life expectancy has become more important nowadays as life has become more vulnerable due to many factors, including social, economic, environmental, education, lifestyle, and health condition. A lot of studies on life expectancy have been carried out. However, studies focusing on the Asian population are limited. This study presents machine learning algorithms for life expectancy based on the Asian population dataset. Comparisons are made between tree classifier models, namely, J48, Random Tree, and Random Forest. Cross validations with 10 and 20 folds are used. Results show that the highest accuracy is obtained with Random Forest with 84% accuracy with 10-fold cross-validation. This study further identifies the most significant factors that influence life expectancy prediction, which includes socioeconomic factors and educational status, health conditions and infectious disease.