Comparative Analysis of Support Vector Machine, K-Nearest Neighbor, and Multilayer Perceptron for Nutritional Status Classification

Abstrak

Nutritional status plays an important role in describing the balance between nutritional intake and requirements, which can be measured through anthropometric indicators such as weight, height, and body mass index. Nutritional imbalances, whether deficiencies or excesses, can cause various serious health problems. This study aims to conduct a comparative analysis of three machine learning algorithms, namely Support Vector Machine, K-Nearest Neighbor, and Multilayer Perceptron, in classifying the nutritional status of adults based on anthropometric data from the RSUD Kota Banjar. The research stages included data preprocessing (data cleaning, normalization, standardization, and label encoding), dividing the data into train, validation, and test sets, and training the model with the best parameters for each algorithm. The testing was conducted using 10% of the test data, which consisted of 151 samples. The results of the experiment showed that the MLP model provided the best performance with the highest accuracy of 94.04%, followed by SVM and KNN with accuracies of 93.38% each. These results indicate that MLP is superior in recognizing non-linear patterns in anthropometric data and has better generalization capabilities than the other two algorithms. Thus, the MLP model can be used as an effective alternative in machine learning-based nutritional status classification systems to support analysis and early detection of nutritional problems in the health sector.

Identifier

• DOI: https://doi.org/10.5281/zenodo.18280089
• OpenAlex: https://openalex.org/W7124531426