Sentimen Analisis Aplikasi Belajar Online Menggunakan Klasifikasi SVM

Abstract

Google Play Store is where a wide variety of applications are available, whether paid or not. Google Play Store page is a place for application users to express opinions, reviews and ratings. Ruang Guru, Zenius and Quipper are available on the platform. Analysis was carried out using sentiment analysis and SVM algorithm. Data was obtained using data scraping techniques, using help of google-play-scraper library. Web scraping process is divided into 3 stages namely Fetching, Extraction, and Transformation. Data collected is 30,000 data which is divided into 10,000 data for each application. Research begins with data preprocessing stage which includes normalization, case folding, cleaning, tokenizing, and stopwords. then data is divided into 90% training data and 10% test data. Training data is labeled with values 1, 0, and -1. Value 1 means positive, value 0 means neutral and -1 means negative. Results of classification sentiment using SVM show that Ruang Guru has highest positive value compared to Zenius and Quipper. However, user response equally gives a positive value for application. Accuracy value of research shows that sentiment classification data with SVM has an average accuracy for Ruang Guru of 99%, Zenius of 96%, and Quipper of 82%.