Sentiment Analysis Of Indosat's Mobile Operator Services On Twitter Using The Naïve Bayes Algorithm

E-ISSN : 2807-9035 Volume 4, Number 1, May 2024 https://doi.org/10.47709/brilliance.v4i1.4084 Sentiment Analysis Of Indosat's Mobile Operator Services On Twitter Using The Naïve Bayes Algorithm Sufajar Butsianto1*, Sifa Fauziah2, Candra Naya3, Futuh Maulana4 1,2,3,4 Informatics Engineering Study Program, Faculty of Engineering, Pelita Bangsa University, Indonesian , , , 4 1 *Corresponding Author Article History: Submitted: 12-06-2024 Accepted: 13-06-2024 Published: 28-06-2024 Keywords: Twitter, Data Mining, Naïve Bayes, Rstudio Brilliance: Research of Artificial Intelligence is licensed under a Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). ABSTRACT Twitter is a social media that allows users to share information with others in real time. Information that is shared on Twitter is usually referred to as a tweet. Sentiment analysis is a branch of research in the text mining domain where the process of identifying and extracting sentiment data will usually be categorized based on its polarity, whether it is positive, negative or neutral. We can process data from opinions on Twitter using data mining techniques, namely classification. The algorithm that will be used in this research is the Naïve Bayes Algorithm. This research will also use the RStudio application. It is a computer programming language that allows users to program algorithms and use tools that have been developed through R by other users. R is a high-level programming language and is also an environment for data and graph analysis. Based on the experimental results, using a comparison of training data and test data of 20%: 80%, 40%: 60%, 60%: 40%, 80%: 20% and 90%:10%, the results of sentiment classification using the Naïve Bayes method are obtained. and using 10-fold cross validation obtained an average value of 85.00% accuracy and The decrease in machine learning performance occurs in the ratio of 80:20 or 1440 training data: 360 data testing, while the ratio of 20%:80% and 90%:10% has the same accuracy value, namely 85.41%. INTRODUCTION The development of the world of technology and information is increasingly moving in a digital direction day by day. The digital era has made humans enter a new lifestyle that cannot be separated from electronic devices. Technology is a tool that helps human needs, with technology everything can be done more easily. Challenges in the digital era have also entered various fields such as politics, economics, social culture, defense, security and information technology itself. The digital era was born with the emergence of digital, internet networks, especially computer information technology. The new media of the digital era has the characteristics of being able to be manipulated, being network or internet in nature. . The media capabilities of this digital era make it easier for people to receive information more quickly.(Megawati, 2021) The role of technology is so important that it is starting to bring civilization into the digital era and increasingly rapidly and the current development of communication technology has changed people's habitual tendencies in expressing their opinions on social media. One of the social media that is popular among internet users today is Twitter. The most popular social media for expressing opinions is Twitter (Syarifuddin, 2020). Twitter is a social media that prioritizes socializing using text, although the new version supports video and photo formats to support tweets. In this way, Twitter is the right tool for collecting public sentiment data on the internet.(Asro’i & Februariyanti, 2022) Twitter is a social media that allows users to share information with other people in real time. Information shared on Twitter is usually called a tweet (Twett).(Elsa Annisa Batu Bara et al., 22 C.E.). Indonesia is in fifth position as the country with the most Twitter users after England and other large countries (Kominfo.go.id, 2020). With so many social media users, social media marketing has become a strategic key in marketing activities in the world. This is proven by 94% of companies in the world using social media for marketing purposes. Likewise, cellular operator companies use Twitter as a promotional medium, one of which is Indosat. Indosat currently has 1.5 million followers with official usernames. Indosat is one of the largest cellular operators in Indonesia. However, Indosat also does not escape the comments of its users with various kinds of positive and negative comments. The large number of Indosat cellular operator users who submit comments can be used to search for information to analyze comments on Twitter using sentiment analysis.(Syailendra Reza Irwansyah Rezeki et al., 2020) Sentiment analysis is a branch of research in the text mining domain where the process of identifying and extracting sentiment data is usually categorized based on its polarity, whether positive, negative or neutral. We can process opinion data on Twitter using data mining techniques, namely classification. To group sentiment, the author divides 3 indicators, namely positive, negative and neutral sentiment with indicators based on tweets. Currently Twitter This is an Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. 245 E-ISSN : 2807-9035 Volume 4, Number 1, May 2024 https://doi.org/10.47709/brilliance.v4i1.4084 is a good indicator for influencing research. This sentiment analysis is carried out to determine public sentiment about something using a machine learning approach (Putranti & Winarko, 2014) LITERATURE REVIEW Sentiment Analysis on Twitter Regarding Post-Disaster Using the Naïve Bayes Method with the N-Gram Feature(Rozi et al., 2023), explained that the Naïve Bayes Classifier Algorithm can be used to classify tweets into positive or negative, especially tweets regarding post-disaster . And testing the accuracy of the algorithm which was carried out by manually labeling 15 respondents, it was found that the results from unigrams and bigrams had quite significant differences. From these four tests, the highest accuracy results were obtained for unigrams, namely 93.33% and bigrams, 86.67%. Classification of public opinion towards the MyRepublicId ISP service using a dataset on Twitter and carried out by applying the naïve Bayes method produces accuracy in the categories of positive 0.976%, neutral 0.833%, and negative 0.82895%, with an average value of 0.87949%, the theory is presented by Hafiz Irsyad, Ahmad Farisi, Muhammad Rizky Pribadi in his research entitled "Classification of Public Opinion on MyRepublic ISP Services using Naive Bayes"(Hafiz Irsyad et al., 2019) Dedi Darwis(Darwis et al., 2021) in their research "Application of the Naive Bayes Algorithm for Sentiment Analysis Review of National BMKG Data" explains that the process of extracting data from National BMKG Twitter uses the Python 3.74 programming language with preprocessi (...truncated)