Pengaruh Metode Reduced Error Pruning pada Algoritma C4.5 untuk Prediksi Penyakit Diabetes

Abstract

Degenerative disease is one of the conditions that can cause the performance of several organs in the human body to decrease and affect health conditions. The prevalence rate of diabetes is predicted to continue to increase in 2030 to reach 578 million and 700 million in 2045. In this research, a diabetes prediction system was formed using the C4.5 Algorithm with Reduced Error Pruning (REP). This research is focused on the application of the Reduced Error Pruning method on the C4.5 Algorithm and used two datasets containing several medical predictors of diabetes symptoms. Based on the research that has been done, the prediction process using the C4.5 Algorithm with Reduced Error Pruning based on the first dataset resulted in an average accuracy of 92,4% with an average accuracy before Reduced Error Pruning of 91,6%. In comparison, in the second dataset, average accuracy was obtained without Reduced Error Pruning by 81,2% and 83,4% for results with Reduced Error Pruning. Based on this percentage, the Reduced Error Pruning method does not have a big influence on the level of accuracy produced.