HYBRID METODE BOOSTRAP DAN TEKNIK IMPUTASI PADA METODE C4-5 UNTUK PREDIKSI PENYAKIT GINJAL KRONIS

Abstract

Missing values is a serious problem that most often found in real data today. The C4.5 method is a popular classification predictive modeling used because of its ease of implementation. However, C4.5 is still weak when testing data that contains large missing. In this study we used a hybrid approach the bootstrap method and k-NN imputation to overcome missing values. The proposed method tested using Chronic Kidney Disease (CKD) data, and evaluated using accuracy and AUC. The results showed that the proposed method was superior in overcoming missing values in CKD. It can be concluded that the proposed method is able to overcome missing values for chronic kidney disease prediction.