Komparasi Algoritma Naive Bayes dan K-Nearest Neighbor untuk Membangun Pengetahuan Diagnosa Penyakit Diabetes

Abstract

Diabetes is caused by a deficiency of the hormone insulin, which is secreted by the pancreas to lower blood sugar levels. The factors that trigger the occurrence of diabetes are derived from various factors such as a combination of genetic and environmental factors. The phenomenon of the emergence of various beverage brand outlets can be one of the triggers for blood sugar levels in humans. Normal blood sugar levels in the body range from 70-130 mg/dL before eating, less than 180 mg/dL two hours after eating, less than 100 mg/dL after not eating or surviving for eight hours, and 100-140 mg/dL at bedtime. This research aims to determine which algorithm is suitable for building knowledge about diabetes using the Naïve Bayes and K-Nearest Neighbor (KNN) algorithm. The accuracy results from Naïve Bayes are 85.60% and K- Nearest Neighbor of 91.61%. The results showed that K-Nearest Neighbor proved to have the best accuracy.