Implementasi Algoritma K-Nearest Neighbor (KNN) Untuk Prediksi Penyakit Diabetes Berbasis Web
TL;DR
A web application for predicting diabetes risk using the K-Nearest Neighbor (KNN) classification algorithm and results indicate that the developed application can support early diabetes-risk screening and demonstrate that KNN can be integrated into a web-based prediction system.
Abstract
Diabetes mellitus is a chronic metabolic disease that requires early detection to reduce the risk of serious complications. Limited access to practical screening tools encourages the development of a web-based prediction system that can be accessed independently. This study develops a web application for predicting diabetes risk using the K-Nearest Neighbor (KNN) classification algorithm. The system processes health parameters consisting of glucose level, blood pressure, body mass index (BMI), age, and gender. System development applies the Software Development Life Cycle (SDLC) Waterfall model, covering requirements analysis, system design, development, testing, and maintenance. The developed application provides data management, K-value configuration, prediction, and prediction-history features. Black Box testing shows that the implemented functions operate according to the expected scenarios. Accuracy testing using 10 test data produced 10 correct predictions and no incorrect predictions, resulting in an accuracy of 100%. The results indicate that the developed application can support early diabetes-risk screening and demonstrate that KNN can be integrated into a web-based prediction system.
