<![CDATA[Social media platforms like YouTube are frequently used by the public and can quickly make issues go viral. Recently, customs duties have come under scrutiny for being considered too high. For example, a man who bought shoes worth 10 million rupiahs was charged 30 million rupiahs in import duties, and a female migrant worker from Madura was charged hundreds of millions of rupiahs for bringing 3kg of gold from Saudi Arabia. These cases have sparked public debate, leading to a sentiment analysis using the Naïve Bayes algorithm and SMOTE method. The research dataset was imbalanced, prompting a comparison between using SMOTE and not using it. The evaluation results without SMOTE showed an accuracy of 95.175%, with precision at 95%, recall at 100%, and an F1-score of 98% for the negative class, but all metrics for the positive class were 0%. After applying SMOTE, the overall accuracy was 85.526%. The negative class achieved a precision of 98%, recall of 87%, and an F1-score of 92%, while the positive class achieved a precision of 19%, recall of 64%, and an F1-score of 30%. Without SMOTE, the accuracy was higher, but overfitting occurred.]]>
Copyrights © 2024
