Perbandingan Support Vector Machine dan Random Forest dalam Analisis Sentimen Komentar YouTube Terkait Isu Hak Veto Amerika Serikat

Abstract

This study aims to compare the performance of two classification algorithms Random Forest and Support Vector Machine (SVM) with a sigmoid kernel in conducting sentiment analysis on YouTube comments related to the issue of the United States’ veto power. The dataset consists of 3,363 comments that have undergone comprehensive preprocessing steps (cleaning, normalization, tokenization, etc.) and were manually labeled into two sentiment classes: positive and negative. The findings indicate that SVM provides a more balanced classification across both sentiment categories, although its overall accuracy is slightly lower at 88.00%. In contrast, Random Forest achieves the highest accuracy at 89.33%, making it superior in terms of overall predictive performance. Therefore, SVM is more suitable when balanced class performance is the priority, whereas Random Forest is preferable when maximizing classification accuracy is the primary objective.