<![CDATA[One of the primary health issues that significantly affects children's growth and development and their future economic potential is stunting in toddlers. To prevent long-term impacts, identification of stunting status needs to be done early so that nutritional and health interventions can be provided effectively.. This research aims to identify the stunting status in toddlers using the K-Nearest Neighbor (K-NN) method. The dataset used includes various attributes of toddlers, such as gender, age, weight, height, weight-for-age index (W/A), z-score of weight-for-age index (W/A), weight-for-height index (W/H), z-score of weight-for-height index (W/H), height-for-age index (H/A), and z-score of height-for-age index (H/A). The data is then divided into training (80%) and testing (20%) sets. The K-NN model is trained with the training data and tested on the testing data. The initial evaluation of the model shows that the K-NN model has an accuracy of 91.20%, precision of 85%, recall of 28%, and an F1 score of 42%, indicating the effectiveness of this method in identifying the stunting status in toddlers, despite some shortcomings. To address these shortcomings, the K-NN model is combined with the forward selection method, which significantly improves the model's performance, achieving an accuracy, precision, recall, and F1 score of 100%. This combination demonstrates that K-NN with forward selection is highly effective and can produce a more accurate and reliable model.]]>
Copyrights © 2024
