Analisis Sentimen Aplikasi WETV di Google Play Store Menggunakan Algoritma Support Vector Machine

Abstract

WeTV is an online streaming application widely used by Indonesia’s people as an entertainment medium while at home. This application has been downloaded more than 50 million times on the official Google Play Store website. The number of users who use it makes the reviews of this application abundant as well. Large numbers of reviews are very difficult to read manually, sentiment analysis is needed to classify reviews into positive and negative classes. This study uses a support vector machine algorithm with a linear kernel to classify review data from the WeTV application. KDD was used as a method to complete this research. In the analysis process to obtain information, 4 scenarios were carried out, with the division in the first scenario consisting of 60% training data and 40% test data, the second scenario consisting of 70% training data and 30% test data, the third scenario 80% training data and 20% test data, and the last scenario 90% training data and 10% test data. The highest test results of 85% were obtained from the second scenario with the distribution of training data of 70% and 30% of test data, the third with the distribution of training data of 80% and 20% of test data, and the fourth with the distribution of training data of 90% and test 10% data. The confusion matrix is used as an evaluation of the model that has been made, the results show an accuracy in the first scenario of 83%, with a precision value of 83%, recall 89%, and an f1-score of 86%. The accuracy in the second scenario is 85%, precision is 86%, recall is 89%, f1-score is 87%, accuracy in the third scenario is 85%, precision is 85%, recall is 90%, and f1-score is 88%. And the fourth scenario gets an accuracy of 85%, precision 86%, recall 90%, and f1-score 90%.