ANALISIS SENTIMEN OPINI PUBLIK PADA MEDIA DIGITAL TERHADAP PERNYATAAN PRABOWO TENTANG NILAI TUKAR RUPIAH MENGGUNAKAN SUPPORT VECTOR MACHINE DENGAN SELEKSI FITUR CHI-SQUARE
Abstract
ABSTRACT
The development of digital media has encouraged people to become increasingly active in expressing their opinions on various political and economic issues through social media, one of which is YouTube. Prabowo Subianto's statement regarding the rupiah exchange rate generated various public responses, as reflected in the YouTube comment section. The large number of comments makes manual analysis less effective, requiring an automated method to classify sentiment. This study aims to analyze public opinion sentiment regarding Prabowo's statement on the rupiah exchange rate using the Support Vector Machine (SVM) algorithm with Chi-Square feature selection. The research data consisted of 1,087 comments from a Kompas TV video collected using YouTube Comment Downloader. The research stages included data selection, preprocessing consisting of case folding, cleaning, tokenizing, stopword removal, and word normalization, manual sentiment labeling, dataset splitting with an 80:20 ratio, TF-IDF weighting, Chi-Square feature selection, and classification using the SVM algorithm with the Scikit-Learn library. Based on the revised labeling and review of the comments, the dataset consisted of 138 positive comments and 949 negative comments. The testing results showed that the SVM model with Chi-Square feature selection achieved an accuracy of 86.24% with a weighted F1-score of 0.86. These results indicate that the model was able to classify sentiment in Indonesian-language YouTube comments, although its ability to identify the positive class was lower than its ability to identify the negative class.
Keywords: Sentiment Analysis, Support Vector Machine, Chi-Square, TF-IDF, Digital Media, YouTube.

