Opini Public Terhadap Liga Korupsi Indonesia Pada Platform Youtube menggunakan Naive Bayes dan SMOTE

Abstract

This study analyzes public opinion regarding the Corruption League in Indonesia by utilizing the Naïve Bayes method combined with the Synthetic Minority Oversampling Technique (SMOTE). The Corruption League is a compilation of corruption cases involving public officials, politicians, and other parties in Indonesia. In this research, Naïve Bayes is employed for sentiment classification, while SMOTE is used to address class imbalance within the dataset, which was collected from YouTube comments. The methodology consists of several stages, including data collection, labeling, preprocessing, classification, and model evaluation. The results reveal that Naïve Bayes without SMOTE achieves high performance in identifying the negative class but struggles significantly in recognizing the positive class, leading to an imbalanced classification outcome. Conversely, when Naïve Bayes is combined with SMOTE, the model’s performance becomes more balanced, showing a notable improvement in detecting the positive class. Additionally, accuracy increases from 79.7% to 84.3%. This study provides valuable insights into public perceptions and demonstrates the effectiveness of classification methods in the context of corruption issues in Indonesia.