Analysis of Sentiment Adiraku App Reviews on Google Play Store Using Vector Machine Support Algorithm and Naïve Bayes

Abstract

The Adiraku application is considered to be able to facilitate and facilitate customers so that there is no need to come to the branch office to get information related to the number of installments that must be paid, due dates, credit simulations, and Adira Finance information offers to customers. A large number of reviews from users received makes it difficult for developers to read them, it will take too much time and effort if they have to read and analyze them manually. To find out which reviews are classified as positive or negative reviews. need a sentiment analysis of the review. This study aims to find out how the opinions or opinions of its users on the services of the application, by analyzing these sentiments through a classification process using two algorithms, namely Support Vector Machine and Naïve Bayes. The data used amounted to 2000 data obtained from Google Playstore. Data is labeled into 2 classes namely positive class and negative. Furthermore, the data is divided into 70% training data and 30% testing data and methods used for testing using Bernoulli Naïve Bayes and Linear Kernel. It was concluded that the number of user reviews of the Adiraku application on the Google Play Store showed more positive comments, amounting to 1412 positive and negative reviews, which was 588 reviews. The Support Vector Machine algorithm performs better by getting the best accuracy value of 96%, while the Naïve Bayes algorithm gets an accuracy value of 85%.