Obesity Prediction: K-Nearest Neighbor vs. Support Vector Machine

Obesity, a significant driver of non-communicable diseases and mortality, presents a growing health concern, with Indonesia witnessing a concerning rise in prevalence. This study compares the efficacy of K-Nearest Neighbor and Support Vector Machine algorithms in predicting obesity, aiming to identify the superior approach. Utilizing a dataset of 1822 patient records with 17 attributes, a 70/30 train-test split was employed within the Orange Data Mining framework. Our findings demonstrate that KNN achieved a higher accuracy level of 98.2% compared to SVM's 94.3%. Furthermore, KNN exhibited a lower error rate in the confusion matrix analysis, indicating its superior performance in accurately classifying obesity cases. This suggests that KNN holds promise as a more reliable method for obesity prediction, potentially aiding early detection and intervention strategies.