Comparison of Chaid, Cart, and Random Forest in Classification of Anemia and Non-Anemia in Young Women in Indonesia

Abstract: Good health and well-being is one of Sustainable Development Goals (SDGs) indicators that needs to be achieved to create a quality resource. One of the health problems that often occurs in Indonesia is anemia. Anemia is a condition where the hemoglobin (Hb) level is less than normal, which is less than 12,0 g/L. Menstruation, growth, sexual maturity and lack of iron intake due to wrong diet patterns could cause young women to be susceptible to anemia which will affect intelligence and comprehension. Research on the factors that affect anemia in young women needs to be conducted. CHAID, CART and random forest are methods that can classify factors that affect anemia. This study aims to use these three methods to analyze the factors that influence anemia in young women and investigate the best method based on good measure of sensitivity, specificity, and accuracy. The results of the analysis of CHAID, CART and random forest showed that there were 2 variables that both played a role in separating anemia from non-anemia in young women, namely pregnant status and nutritional status. The CHAID method is the best model for classifying anemia and non-anemia in young women that has a sensitivity value of 81.43%.