Optimasi Klasifikasi Stunting Balita dengan Teknik Boosting pada Decision Tree

Abstract

Malnutrition in the growth of small children is known as stunting. Currently, nutrition is still a serious problem that needs to be addressed, especially the nutrition of children under five. Considering the target prevalence rate (14%) in 2024 and how dangerous stunting is in Indonesia, this stunting problem needs to be addressed. The purpose of this research is to optimize the decision tree algorithm in stunting classification using boosting technique optimization. The boosting techniques used are AdaBoost, XGBoost, and Gradient Boosting methods. The boosting technique was chosen because it can improve classifier performance by combining multiple models that are learned sequentially, resulting in more effective predictions. This research uses infant data from Kaggle, which has a total of 10,000 data points, 8 attributes, and 2 classes. Based on the results of this study, decision tree optimization using the XGBoost method achieved the best results with accuracy of 83.8%, precision of 82%, recall of 83.8%, and F1-score of 81.2%, which shows great potential in improving the classification of stunted infants. The boosting technique is the best choice compared to other techniques. Based on the results of this study, the boosting technique can accurately predict and demonstrate a high level of precision in handling stunting classification.