It is important for doctors to make an early diagnosis of tuberculosis in order to reduce the transmission of the disease to the wider community. In this study, the authors will apply and compare several methods of data mining classification, including AlgoritmaC4.5, Naïve Bayes, and Neural Network to diagnose tuberculosis disease, then compare which of the three methods are the most accurate. Based on the performance measurement results of the three models using Cross Validation, Confusion Matrix and ROC Curve methods, it is known that Naïve Bayes method is the best method with accuracy of 94.18% and under the curva (AUC) value of 0.977 , then neural network method with accuracy 89,89% and under the curva value (AUC) 0,975, and then C4.5 method with accuracy level equal to 84,56% and under the curva value (AUC) equal to 0,938. This shows that the three models that are produced including the category of classification is very good because it has an AUC value between 0.90-1.00.
Copyrights © 2018