| dc.description.abstract | Children under five years are the most vulnerable age group in terms of nutrition and health in a community. Nutritional problems often occur at the age of five, but the diagnosis of malnutrition is still done by directly measuring nutrition indicators such as weight, height or biochemical markers of some nutrients. With the technological advances in computing and statistical data processing program, the available non-nutritional data can be used to predict the nutritional status of children. The objective of this study was to investigate the use of support vector regression (SVR) as a machine-learning method to find models that can predict the nutritional status of children and to develop prediction system from the SVR models. The best model was produced by RBF kernel, with the highest degree of correlation and the lowest error in each type of Z-score predicted. With the best SVR model, a system that can predict Z-score can be developed, although it is not quite accurate. | en |
