Supervised machine learning approaches to predict infant mortality: a case study of the 2014 Kenya Demographic and Health Survey

Abstract

The study used supervised machine learning approaches to predict infant mortality in Kenya and the 2014 Kenya Demographic and Health Survey. Di erent classi cation methods were used. The methods were Logistic regression, K-nearest neighbor and Random forest model. Random Forest performed well with an accuracy of approximately 97.1% followed by Logistic Regression model with 86.1% and K-nearest neighbor with 85.6%. The results concluded that random forest model was the best performing model.