Analisis Sentimen Opini Pelanggan Aplikasi Pln Mobile Menggunakan Metode Vader Lexicon & Naive Bayes

Abstract

The PLN Mobile app was launched by PLN as a superior digital platform to meet all customer needs, provide convenience, and provide a different electricity service experience. With a Google PlayStore rating of 4.8 out of 5, the PLN Mobile app provides ease and speed of service to PLN customers, with approximately 24.8 million users and over 35 million registered customer IDs (https:/web.pln.co.id/media/press-release/2022/07, July 2, 2022).User reviews on the Google Play Store app have a rating scale from 1 to 5. Sometimes users provide ratings that don't match their reviews, thus failing to adequately describe the quality of the app. The number of reviews on the PLN Mobile app is so huge that reading them all will take time. The ranking is used to determine public opinion. The sentiment analysis makes use of 1,000 review sample data from the PLN Mobile app collected between January and June 2022.The initial stages of the research consisted of collecting review data using web scraping, machine translation, data tagging, text preprocessing (TF-IDF), text classification, and model evaluation techniques. For the Lexicon-based text classification approach, with the Vader Lexicon dictionary-based approach, the tagging results were 489 positive opinions, 145 negative opinions, and 366 neutral opinions. Based on the results of a comparison of the positive, neutral, and negative classes of sample data with reviews based on user inappropriateness ratings, the positive class received a 67% rating, the neutral class received a 6% rating, and the negative class received a 27% rating. In addition, the Naive Bayes method is used in the classification process. The author uses a 90:10 split data ratio for the distribution of test and training data. The confusion matrix evaluation process produces an accuracy of 70%.