Penerapan Metode Support Vector Machine (SVM) dalam Klasifikasi Penderita Diabetes Mellitus

Abstract

Diabetes Mellitus (DM) is a chronic disease characterized by the body's inability to metabolize carbohydrates, fats, and proteins, resulting in increased blood sugar (hyperglycemia) due to low insulin levels. Diabetes is due to a combination of heredity (genetics) and unhealthy lifestyles. Hemoglobin A1c is a blood test used to diagnose and manage diabetes patients when measuring blood sugar levels. This study aims to analyze predictive models for the classification of people with diabetes using R Shiny and evaluate the results of the support vector machine method's classification performance. There are many ways to diagnose diabetes, and the support vector machine is one of the machine learning algorithms used in this study's classification case (SVM). This study uses data from Diabetes 130-US Hospital For Years 1999-2008, which was sourced from the UCI Machine Learning Repository and consists of 34 variables and 84900 records, with dataset distribution and testing techniques using the 10-fold cross-validation method and three kernels in modeling using SVM, namely linear, Gaussian, and polynomial. The results obtained are a simple predictive model analysis system for classifying people with diabetes with shiny, making it easier for users to find out the prediction results and obtain the highest accuracy result, which is 82.76 percent of the gaussian kernel.