Application of PCA and K-Means Clustering Methods to Identify Diabetes Mellitus Patient Groups Based on Risk Factors

Abstract

Diabetes mellitus is a chronic disease characterized by high levels of glucose (sugar) in the blood that is high for a long period of time. Identification is the process of recognizing and determining the characteristics of a particular object or entity. hypertension (high blood pressure), smoking and lack of physical activity can affect the condition of diabetes mellitus patients. Therefore, an approach is needed that can identify groups of diabetic patients based on their risk factors, so that appropriate management and treatment can be carried out. The purpose of this study is to apply PCA method by reducing data dimension to identify the linear combination of the most contributing risk factors in diabetes mellitus patient data and apply K-Means Clustering to cluster into groups based on similar risk factors. The methods to be used are Principal Component Analysis (PCA) and K-Means Clustering. type of quantitative research, this research can be categorized as analytic research, variables are risk factors for diabetes mellitus disease. The results of research using the PCA (principal component analysis) method obtained 9 main components (PC) 86.9275%. correlation between attributes and principal components, then a matrix component is formed with a loading value that the greater the value, the stronger the correlation with the principal component formed with a cut off point of loading value> 0.4 regardless of positive and negative. By using the K-Means Clustering method, The clustering results obtained are divided into 3 groups of diabetes patients based on existing risk factors. Centroid C1 represents a group of diabetes mellitus patients whose condition is at a mild level, while Centroid C2 represents a group of diabetes mellitus patients who are at a moderate level, and Centroid C3 represents a group of patients with severe or dangerous diabetes mellitus.