<![CDATA[In data mining, a known Classification model that can be used to identify thyroid disease is Naive Bayes and Bayes Network methods. In this study, a model is made by using both algorithms. the data used are taken from the data of Patients with thyroid by using the tools KNIME. The model then compared to determine the best algorithm in the determination of disease identification. To measure the performance of the two algorithms, it used methods of testing of cross-validation and split percentage. The measurement results using confusion matrix and ROC curves. By using the confusion matrix, Bayes Network has higher accuracy with 98,491% compared with the Naive Bayes with 91,803%. Using the ROC curve, Bayes Network also has higher accuracy with the ROC curve - negative (0.9337), ROC - hyperthyroid (0.9933) and ROC - hypothyroid (0.9977). while Naive Bayes with ROC curve - negative (0.8760), ROC - hyperthyroid (0.9789) and ROC - hypothyroid (0.9018). The method which has very good classification is sequentially Bayes network and naïve Bayes based on assessment AUC between 0.90-1.00. thus the Bayes Network algorithm can provide solutions to the problems of identifying thyroid disease.]]>
Copyrights © 2018
