Algoritma Extreme Gradient Boosting (XGBoost) dan Adaptive Boosting (AdaBoost) Untuk Klasifikasi Penyakit Tiroid

Abstract

Thyroid disease is a disease of the thyroid gland that can interfere with daily activities. Early detection of thyroid disease can have an important impact in optimizing the development of early detection systems that are more effective and accurate in detecting the disease. Data mining approaches can be used to solve this problem by utilizing various available algorithms, such as Adaptive Boosting and Extreme Gradient Boosting. This research aims to improve the development of early thyroid disease prediction by comparing the two algorithms by utilizing the percentage split method.  This research provides results if the Adaptive Boosting algorithm provides an accuracy value of 97%. In class 0, the precision and recall values are the same at 98%, while in class 1 it is 80% and 90%. Meanwhile, testing using the Extreme Gradient Boosting algorithm gives an accuracy value of 98%. In class 0, the same precision and recall values are 99%, while for class 1 it is 86% and 90%. Based on the comparison by considering the accuracy, precision, and recall values, as well as the performance of the two algorithms, it is concluded that the implementation of the Extreme Gradient Boosting algorithm has the best performance for thyroid disease detection.