Shahrood University of Technology Journal of AI and Data Mining 2322-5211 12 4 2024 11 01 Comparative Analysis of Tree-Based Machine Learning Algorithms on Thyroid Disease Prediction Using ROS Technique and Hyperparameter Optimization 511 520 3360 10.22044/jadm.2025.14990.2598 EN Elahe Moradi Department of Electrical and Computer Engineering, Yadegar-e-Imam Khomeini (RAH) Shahre Rey Branch, Islamic Azad University, Tehran, Iran. Journal Article 2024 08 27 Thyroid disease is common worldwide and early diagnosis plays an important role in effective treatment and management. Utilizing machine learning techniques is vital in thyroid disease diagnosis. This research proposes tree-based machine learning algorithms using hyperparameter optimization techniques to predict thyroid disease. The thyroid disease dataset from the UCI Repository is benchmarked to evaluate the performance of the proposed algorithms. After data preprocessing and normalization steps, data balancing has been applied to the data using the random oversampling (ROS) technique. Also, two methods of grid search (GS) and random search (RS) have been employed to optimize hyperparameters. Finally, employing Python software, various criteria were used to evaluate the performance of proposed algorithms such as decision tree, random forest, AdaBoost, and extreme gradient boosting. The results of the simulations indicate that the Extreme Gradient Boosting (XGB) algorithm with the grid search method outperforms all the other algorithms, obtaining an impressive accuracy, AUC, sensitivity, precision, and MCC of 99.39%, 99.97%, 98.85%, 99.40%, 98.79%, respectively. These results demonstrated the potential of the proposed method for accurately predicting thyroid disease. https://jad.shahroodut.ac.ir/article_3360_38ea7c234d9cbb1c8ad17689c559bf6e.pdf