Prediksi Stunting Menggunakan Algoritma Decision Tree Berbasis Synthetic Minority Over-sampling Technique (SMOTE)

Abstract

Stunting is a sign of a serious malnutrition problem and can cause toddlers to have a short height and affect the growth and development of toddlers. The data in the study used anthropometric data of toddlers from the Majasem Health Center in September 2024, a total of 1368 toddler data that had been recorded in that time span. The Decision Tree model used aims to predict the status of toddlers, based on the stunting category. However, the Decision Tree algorithm often faces the problem of non-optimal accuracy due to the imbalance of class data in the dataset. The SMOTE method is used as an effort to overcome the problem of class imbalance, so that the classes in the dataset will be balanced. This research successfully proves that the SMOTE method is able to improve the accuracy of the Decision Tree model, the accuracy of the model with the best SMOTE is 98.56% on training data and 96.91% on testing data with a data proportion of 80:20. This research is useful for helping health workers to provide better insight into the nutritional status of toddlers, besides that the developed model can increase the effectiveness of public health interventions.