Classification of diabetes mellitus disease at Rato Ebuh Hospital-Indonesia using the K-Nearest neighbors method based on missing value

Abstract

Diabetes mellitus is a chronic disease often caused by high blood glucose levels and insufficient insulin production. This research aims to address the classification problem of diabetes mellitus using the K-Nearest Neighbor (K-NN) method. The aim of this research is to create a machine learning model that can detect diabetes early. The study was conducted at Syarifah Ambami Rato Ebu Hospital in Bangkalan, utilizing data from 120 patients in 2019, employing data mining techniques to classify diabetes mellitus patients. Additionally, the steps in data mining involve determining significant variables or features for classification Cleansing and normalization and transformation. The research compares training test results with ratios of 90:10, 80:20, and 70:30. Experimental results show that K-NN with a neighbor value of K=11 achieves the highest accuracy rate of 83% a reduced error rate of 16.67%, and the highest AUC value of 0.7407. These results indicate that the 90:10 data split ratio yields the best model performance in terms of accuracy and class differentiation for diabetes mellitus, as well as the lowest error rate compared to other data split ratios. This study provides a better understanding of diabetes mellitus and demonstrates that K-NN is effective in addressing classification problems, focusing on specific variables that influence the disease. Therefore, it can be concluded that K-Nearest Neighbor (K-NN) is a suitable algorithm for classifying diabetes mellitus.