Comparison of CART and CART ARCING to Classify Household Food Security of Tuberculosis Patients in Surabaya

Abstract

Indonesia shall fulfill the residents need of food by its heterogeneous and big resource of food. But in fact, Indonesian food security is still far from expectation because of the maintenance of food resource in Indonesia has not been maximized yet. Meanwhile, households with tuberculosis patients in Surabaya tend to classify as food-insecure. This study conduct a classification of households with tuberculosis patients food security in Surabaya by using Classification and Regression Tree (CART) and Adaptive Resampling and Combining (ARCING). The purpose of this study is to know the influential factor of households food security and the accuracy of its classification. As the result, the way households obtain foods is the most influential factor of Indonesian food security in Surabaya. Other than that, this study shows that accuracy when using ARCING is higher than CART.