Stunting is a serious problem caused by chronic malnutrition in children under five, causing stunted growth and having a negative impact on long-term health and productivity. Therefore, early detection of stunting is very important to reduce its negative impacts. Previous studies utilizing machine learning have proven the success of this method in various health applications, such as disease detection and the prediction of medical conditions. The results of this research are a comparative evaluation of five classifications, namely Logistic Regression, Decision Tree, Random Forest, K-Nearest Neighbors (KNN), and Support Vector Machine (SVM), in classifying stunted toddlers. The dataset used contains four important attributes: age, gender, weight, and height of toddlers, as well as a binary class label that differentiates between toddlers who are stunted and those who are not. The evaluation results show that KNN at K = 3 produces the highest accuracy of 94.85%, making it the best model for classifying stunting in toddlers. Apart from accuracy, other metrics such as precision, recall, and F1-score are used to analyze the algorithm's ability to solve this problem. KNN stands out as the best model, with the highest F1-score of 89.47%. KNN also manages to maintain a balance between precision and recall, making it an excellent choice for treating stunting in toddlers. Apart from that, the use of the AUC metric from the ROC curve also shows the superiority of KNN in differentiating between stunted and non-stunting toddlers. With a combination of consistent evaluation results, both in terms of accuracy and other evaluation metrics, this research proves that KNN is the best choice for overcoming the task of classifying stunting in toddlers.