Analysis of Customer Satisfaction Sentiment towards PT Pertamina Products and Services on Twitter Using the Naïve Bayes Algorithm

Abstract

PT Pertamina is one company that still uses Twitter. With the existence of Twitter, PT Pertamina can find out by looking at the public's complaints about PT Pertamina's products and services. The number of public comments on Twitter regarding PT Pertamina makes it difficult for PT Pertamina to segment in the form of positive comments, negative comments, and neutral comments. Not only that, PT Pertamina itself also finds it difficult to distinguish between comments related to SPBU (SPBU category) and comments related to SPBE (SPBE category) because on the twitter all comments are combined into one (not categorized by SPBU and SPBE). So that to be able to help obtain this information, segmentation and categorization are needed which should make it easier for PT Pertamina to pay more attention to PT Pertamina itself. The process that can be used to deal with these problems is by using Text Mining to find words that can represent the content that is was in the comments regarding PT Pertamina on Twitter, then classified using the Naive Bayes algorithm. Output of the sentiment analysis system is in the form of positive segmentation, negative segmentation and neutral segmentation. Then there is product categorization based on gas station services and services. So that the results of the segmentation and categories can be a reference for PT Pertamina to improve and improve the quality of products and services to the community. The test results using a k-fold value of 2 to 10 and tested 10 times for each k value resulted in the accuracy of the naïve Bayes algorithm around 99.393% From 627 twitter data collected in the system, it can be seen that public sentiment tends to be positive in the gas station category, around 40.07%. Also, public sentiment tends to be neutral in the SPBE category at around 37.50%.