Analisis Sentimen Opini Publik terhadap Kasus Korupsi Timah di Youtube Menggunakan Metode Oversampling dan Algoritma Decision Tree
Abstract
This study analyzes public opinion regarding the corruption case of PT. Timah (Tbk), which caused state losses of up to IDR 271 trillion, through YouTube comments. The methods used are the Decision Tree algorithm and the SMOTE Oversampling method. A total of 2501 comments were collected and processed. The stages include data preprocessing, sentiment labeling, and model training. The results show that the use of SMOTE improves the accuracy and performance of the model. With SMOTE, the model achieves an accuracy of 56%, a precision of 0.55, a recall of 0.55, and an F1-score of 0.55, while without SMOTE, the model only achieves 54%, a precision of 0.52, a recall of 0.52, and an F1-score of 0.52. Precision, recall, and F1-score also increase when using SMOTE. This study highlights the importance of the Oversampling technique in dealing with class imbalance to improve the accuracy and sentiment analysis model. These results make a significant contribution to sentiment analysis, highlighting the role of SMOTE in overcoming class imbalance and creating a more accurate model.