Comparison of K-Nearest Neighbor Algorithm Performance and Naïve Bayes in Predicting Stroke Disease

Stroke is one of the most dangerous diseases that can cause death and long-term disability. Early identification of stroke risk can help the prevention process. This study compares two classification algorithms, namely K-Nearest Neighbor (K-NN) and Naïve Bayes, in prediction stroke risk based on patient data. The dataset used is 1470 data that has several attributes such as age, hypertension, heart disease, glucose levels, and others. The evaluation results showed that Naïve Bayes algorithm performed better with 73.1% accuracy and 79.9% AUC, compared to K-NN which had 68.4% accuracy and 75.1% AUC. Based on these results, Naïve Bayes algorithm is considered more effective to be used in stroke risk prediction system.