Classification Analysis of Stunting Data on Toddlers in Bekasi City Using Logistic Regression and Random Forest

Stunting in toddlers is one of the most pressing nutritional issues in Indonesia. This condition, caused by chronic malnutrition, is a key indicator of maternal and infant health, particularly during early life. Regular monitoring of a toddler’s growth and development is necessary for preventing stunting. Therefore, developing accurate and efficient predictive models is essential. This study aims to analyze the performance of two Machine Learning models, namely Logistic Regression and Random Forest, in predicting stunting status in toddlers. The dataset used is toddler development data from the Kota Baru Health Center, Bekasi City. Due to the limited amount of data, the data split is performed using k-fold cross-validation with k=5 to maximize the dataset’s utilization. Given the imbalanced data, with 39.4% stunting from the total data, an oversampling method was applied to the minority class. Logistic Regression was selected for its simplicity and interpretability, whereas Random Forest was favored for its ability to handle data complexity by combining predictions from multiple decision trees. The evaluation results indicate that Random Forest outperforms Logistic Regression, achieving an average accuracy of 88.07% and a macro-average F1-score of 87.32%, compared to Logistic Regression’s average accuracy of 85.48% and macro-average F1-score of 84.73%. This research is expected to support stunting prevention in Indonesia by providing a predictive approach based on Machine Learning to improve the monitoring of toddlers’ growth and development.