Sentiment Analysis of Instagram Reviews Using Support Vector Machine and Random Forest Algorithms: A Case Study of Universitas Dian Nusantara

Abstract

The increasing use of social media, particularly Instagram with over 1 billion active users in 2023, creates opportunities for Universitas Dian Nusantara (UNDIRA) to understand public perception through user review analysis. This study aims to develop a machine learning-based sentiment analysis model to categorize UNDIRA Instagram user reviews into positive, negative, and neutral sentiments. The novelty of this study lies in the combined use of TF-IDF and Word2Vec features together with a systematic comparison of SVM and Random Forest on Indonesian-language review data, a context still rarely examined for higher-education institutions. The research involved data collection through Instagram scraping, data preprocessing including stop word removal and stemming, and the application of three machine learning models: Support Vector Machine (SVM) with TF-IDF feature extraction, Random Forest (RF) with Word2Vec, and RF with TF-IDF. Results indicate that SVM with TF-IDF achieved the best performance with 99.11% accuracy and 99.13% F1-Score, outperforming Random Forest at 96.43% accuracy