ANALISIS SENTIMEN MASYARAKAT TERHADAP KASUS JUDI ONLINE MENGGUNAKAN DATA DARI MEDIA SOSIAL X PENDEKATAN NAIVE BAYES DAN SVM

Abstract

Research conducted by analyzing public sentiment related to online gambling cases using datasets from x social media  using the naïve bayes method approach and support vector machine (SVM). The analysis phase starts with data gathering or crawling, followed by data labeling, data preprocessing, and ultimately method categorization. The dataset comprises 2,866 tweets, with 1,436 classified as positive (50.12%) and 1,429 as negative (49.88%). The data before to the classification process is partitioned into training data and testing data, including 70% training data and 30% testing data. The analysis with the SVM approach yielded a classification accuracy of 83%, whereas the naïve Bayes method achieved just 79%. Upon completion of the method classification process, the subsequent phase involves visualization and assessment. During the visualization step, bar plots, word clouds, and word frequencies derived from sentiment analysis calculations are shown, alongside a visualization of words from the dataset. The investigation indicates that the SVM approach outperforms Naive Bayes in sentiment classification. The benefit of SVM resides in its capability to manage data with elevated limits and accuracy, enhancing its efficiency in discerning positive and negative thoughts. The findings of this study demonstrate that SVM is better appropriate for data exhibiting complicated distributions, whereas the Naive Bayes approach yields suboptimal results. Thus, SVM can be proposed as a more appropriate and reliable approach for similar sentiment analysis in the future.