PENERAPAN DATA MINING UNTUK PREDIKSI AWAL KEMUNGKINAN TERINDIKASI DIABETES

  • Erfan Karyadiputra Universitas Islam Kalimantan Muhammad Arysad Al-Banjari Banjarmasin
    (ID)
  • Agus Setiawan
    (ID)

Abstract

Diabetes is one of the chronic diseases that have a characteristic in the form of high levels of blood sugar (glucose). Glucose is the main source of energy for the cells of the human body, but glucose accumulated in the blood can cause various disorders of the body's organs if not controlled and cause various complications of other diseases that can harm the sufferer. Early detection of diabetes is necessary because of the long asymptomatic phase, the asymptomatic phase is a condition of the disease that has been positively suffered but does not cause clinical symptoms in sufferers. The purpose of this study was to predict the initial likelihood of a person being indicated by diabetes based on the diabetes dataset using data mining techniques, namely the Decision Tree C4.5 algorithm method, Naive Bayes and K-Nearest Neighbors. Cross Validation test results from the three algortima were then compared with performance measurements using Confusion Matrix, Compare ROC and Paired T-Test so that the best algorithm method was obtained. The results of this study showed that the Decision Tree C4.5 algorithm became the best algorithm based on the results of predictive accuracy performance of 96.35% with an AUC value of 0.949 so that it belongs to the category of excellent classification and the results of different T-Test tests are dominant when compared to other algorithms. Therefore, the decision tree C4.5 algorithm is more accurate in predicting the beginning of the possibility of someone being indicated by diabetes.

Published
2022-08-23
Abstract viewed = 303 times