Stunting is one of the primary health concerns for children in Indonesia. Preventing stunting in toddlers is essential to mitigate long-term effects on both their health and society as a whole. Preventing stunting involves monitoring the growth of toddlers. Therefore, a predictive system for identifying stunting in toddlers is crucial. Machine learning offers many methods that can be used to build a system to predict stunting conditions in toddlers. This research analyzes some potentially suitable machine learning models for predicting stunting classes using Ensemble Learning, which are Boosted K-Nearest Neighbor (BK) and Boosted Na¨ ?ve Bayes (BN). The boosting is done by assigning an initial weight to each sample and increasing each failed classified sample’s weight. This approach enhances the learning done by the machine learning model by focusing on learning more about the failed classified samples. The dataset has an imbalance issue in this research, with the data categorized as short and very short at less than 2% of the total dataset. Therefore, oversampling of the dataset is done by generating a random dataset based on the distribution of the imbalanced dataset. After that, the normal category dataset is reduced to ensure the data is evenly distributed. The result of elaborating on this oversampling has been unsatisfactory, as the data distribution remains imbalanced despite efforts to stabilize the quantity between classes. Therefore, additional boosting is necessary to ensure proper classification. After the data is balanced by oversampling and boosting, the F-1 score macro average reached 97.44% for the BK method and 57.91% for the BN method. Additionally, the accuracy achieved was 98.62% for BK and 80.62% for BN. These results indicate that the BK method outperforms the BN method, despite the BN method achieving better outcomes than the other previous research.