Sentiment Analysis of Bank BNI User Comments Using the Support Vector Machine Method

The rapid development in the world of information and communication technology has made social media users increase. By looking at various kinds of social media, it is always filled with a variety of service users such as the use of mobile-based banking applications. In Indonesia, almost all banking services use banking facilities such as BNI. By looking at the phenomena that occur in these problems, a study was conducted on comments related to BNI mobile application-based services that are used to improve and update the quality of BNI services to customers so that they can compete with other banks. Thus the researcher aims at classifying the existing BNI Mobile Banking Application user comments on the Google Play service into positive and negative comment sentiment by applying the Support Vector Machine Media method which aims to improve and renew the BNI Mobile Banking Application service system to provide service satisfaction to users BNI. In research conducted using k-fold cross-validation testing obtained SVM kernel linear accuracy values of 78,19% for 60% data training and 40% data testing, meanwhile for 80% data training and 20% data testing get accuracy 76,94% and SVM kernel linear using K-Fold Cross Validation the highest value of 78,45% at 10 fold Cross-Validation. This algorithm has a lightweight computation as evidenced by a dataset of 580 data which only takes 2.5 seconds. K-Fold Cross Validation is proven to be able to optimize a test that was previously worth 78,19% with K-Fold Cross Validation rising to 78,45%