Sentiment Analysis on Ferizy Application Reviews Using Support Vector Machine Method: Analisis Sentimen Pada Ulasan Aplikasi Ferizy Menggunakan Metode Support Vector Machine

Ferizy is a sea transportation ticket booking service that can be accessed by users through official website and  Ferizy application on Google Play Store. To improve the quality of application, Ferizy users can provide reviews of the function and performance of APP through the Google Play Store. Reviews provided by users can be identified as positive or negative sentiment through sentiment analysis with the help of machine learning. In this situation, researchers compared SVM, Naive Bayes, and LSTM methods by covering a dataset of 1,500 reviews. Based on testing, the RBF kernel in this study produces the highest accuracy at a ratio of training data and test data of 90: 10 with accuracy reaching 90.71%. Evaluation of SVM accuracy results using the Confusion Matrix method produces a percentage of 91%. Comparison with Naive Bayes and LSTM methods with the same dataset ratio only produces an accuracy of 83.88% and 85.33%