Sentiment Analysis of Twitter Users Ahead of the 2024 Election Using the Naive Bayes Method

Abstract

Elections are pivotal moments in a democratic nation, where citizens can express their opinions and political preferences. In today's digital era, social media, particularly Twitter, has become a crucial platform for expressing sentiments related to elections. This research aims to analyze Twitter users' sentiments towards the 2024 election using the Naive Bayes method, Knowing the public's views, especially on the Twitter platform, regarding the 2024 election and implementing the Naive Bayes method to classify sentiment. The research method itself consists of data collection from Twitter using the Twitter API, data preprocessing including data cleaning, removal of URLs, hashtags, duplicate words, normalization of words, tokenization, and removing meaningless words using the Rapid Miner application, then testing using training data and testing data in the Naive Bayes method., the data obtained from the keyword "2024 election" on Twitter for the initial data amounted to 2991 data. After going through the cleaning process, clean data amounting to 1069 data was obtained. From the tested data, the results obtained are as follows: The precision class produces an average percentage of true positives of 100.00% while negatives of 81.48%. Class recall produces a percentage of 98.95%, and the accuracy of testing the model is 99.00%. The research results show that the Naive Bayes method has been successfully applied to analyze Twitter user sentiment.