PERBANDINGAN KINERJA METODE K-NEAREST NEIGHBOR (KNN), RANDOM FOREST, DAN DECISION TREE DALAM MEMPREDIKSI DIABETES: Comparing the Accuracy of K-Nearest Neighbour (KNN), Random Forest, and Decision Tree Methods in Predicting Diabetes

Yulianty, Sherly
Najib, Mohamad Khoirun

Al-Aqlu : Jurnal Matematika, Teknik dan Sains
Vol. 3 No. 2 (2025): Juli 2025

Diabetes is a disease with a growing number of sufferers and is the cause of death of 1.5 million people in the world in 2019. A treatment for diabetes is needed, one of which is by predicting diabetics. The K-Nearest Neighbour (KNN), Random Forest, and Decision Tree methods are some methods that can be used to predict diabetes classification. This research aims to compare the performance of KNN, Random Forest, and Decision Tree methods based on accuracy and computation time. The data used in this study are Pregnancies, Glucose, Insulin, Body Mass Index (BMI), and Age as independent variables and Outcome as a dependent variable. The results of research on data that has not been normalised with Min-Max show that the KNN method has a faster computation time than the other two methods, while based on the accuracy value the Decision Tree method has a higher value than the other two methods. Furthermore, the Min-Max normalised data shows a decrease in the accuracy value of the Decision Tree and Random Forest methods, while the accuracy value of the KNN method has increased. Therefore, the Min-Max normalisation treatment is better used for the KNN method.