Classification of Nutritional Status in Toddlers Based on Anthropometric Data Using Random Forest

Abstract

Stunting is a chronic nutritional problem that remains a major challenge in improving child health in Indonesia. This condition has long-term impacts on physical growth, cognitive development, and future productivity of children. Early detection of toddlers' nutritional status is crucial for effectively preventing and addressing stunting cases. This study aims to develop a machine learning-based classification model for toddlers' nutritional status using simple anthropometric data, namely age (in months), sex, and height (in cm). The dataset used in this study was sourced from the 2022 historical records of the Health Department and the Community-Based Nutrition Recording and Reporting System (E-PPGBM), comprising 120,999 entries categorized into four nutritional status classes: normal, tall, stunted, and severely stunted. Data preprocessing included label encoding and feature standardization. The model employed is the Random Forest Classifier, evaluated using accuracy, precision, recall, and F1-score metrics. The training results show that the model achieved a classification accuracy of 99.93% on the test data, with F1-scores for each class as follows: Normal = 0.9998, Severely Stunted = 0.9985, Stunted = 0.9975, Tall = 0.9997. Feature importance analysis indicates that height is the most influential feature in the classification task. These findings demonstrate that machine learning algorithms, particularly Random Forest, are effective for predicting toddlers’ nutritional status and have strong potential to be integrated into digital applications that support Indonesia’s stunting reduction programs. However, the model's limitation lies in its use of only basic anthropometric features—age, sex, and height—without considering additional variables such as weight, disease history, dietary patterns, socioeconomic status, or immunization history, which may also influence a child's nutritional status. To improve the model's accuracy and relevance, it is recommended to incorporate other related features, such as body weight, nutritional intake, health history, and social-economic indicators, in future research.