Performance Analysis of Naive Bayes and Fuzzy K-Nearest Neighbor Methods for Malnutrition Status Classification Systems Devie Rosa Anamisa, Fifin Ayu Mufarroha, Achmad Jauhari, Muhammad Yusuf, Normalita Eka Ariyanti, and Muhammad Hanif Santoso
University of Trunojoyo Madura
Abstract
Malnutrition Status is a nutritional measurement of toddlers^ nutritional needs as indicated by age, weight, upper arm circumference, height, and health status resulting from a balance between daily nutritional needs and intake. Toddlers require adequate nutritional intake in quantity and quality because young children usually have high physical activity levels. Apart from that, efforts to reduce malnutrition status are a top priority in the Health Development program in Indonesia. Therefore, the Government needs a system that can help identify the nutritional status of toddlers early for prevention and treatment based on the classification of their nutritional status, thereby making it easier to collect data on toddlers who experience stunted nutritional status to provide education on increasing stunting nutritional levels. In this study, we classify nutritional status in toddlers by comparing the Naive Bayes (NB) and Fuzzy K-Nearest Neighbor (FKNN) methods. The performance of the two methods was compared to find out which method performed better in classifying malnutrition status. Based on the research results, comparing the performance between the FKNN and NB methods with testing using accuracy as the main benchmark for malnutrition status classification performance. The results showed that the FKNN method was superior in accuracy with a quite large margin of 7.5%. The conclusion is that in classifying toddlers^ nutritional status, the FKNN method outperforms NB.