NUTRITIONAL STATUS INDICATOR OF TODDLERS USING A MACHINE LEARNING APPROACH

Abstrak

Using a machine learning-based methodology that incorporates visualization, classification, and clustering techniques, this study attempts to identify and categorize the nutritional status of toddlers. The dataset, which was obtained from Kaggle, has characteristics like height (in centimeters), gender, and age (in months). The RapidMiner software was used to do the analysis. The K-Means approach was utilized for clustering, while the K-Nearest Neighbor (K-NN) algorithm was utilized for classification. K-NN classification performed well in identifying all nutritional status groups, with an accuracy of 99.93%. From the visualization results using scatter plots and bar charts, it is known that toddlers with the nutritional status of "severely stunted" and "stunted" are mostly experienced by toddlers aged 24 months and above and toddlers with a height of less than 90 cm at the age of over 12 months are most often included in the "stunted" and "severely stunted" categories. K-Means effectively divided the data into four clusters, each of which represented a distinct growth trend according to height and age. The study does not account for other possible affecting factors like weight or socioeconomic position, and is restricted to a dataset with basic attributes (age, gender, and height). By providing a machine learning model for early stunting detection that can be used in healthcare systems and mobile health applications, this study advances the field of health informatics.