Analysis of Public Opinion Sentiment towards the 2024 Presidential Election Based on Clustering Method with K-Means Algorithm

Abstrak

The presence of social media, such as Twitter, Facebook and Instagram, provides a space for people to express their opinions freely and openly. Various sentiments, ranging from support to criticism of the candidates, work programs, and other political issues, have emerged along with the increasing public enthusiasm. Therefore, it is important to understand how public opinion is evolving and what is the main focus of public attention in the 2024 presidential election. The purpose of this research is to analyze the sentiment and views of the public about the presidential election using the Clustering approach and the K-Means method and to classify public opinion for various interests as well as optimizing social media information for the public interest. Based on the research conducted, the K-Means algorithm was successfully applied for sentiment analysis of public opinion on the 2024 presidential election, using tweet data taken through crawling Twitter as many as 220 tweets. From the dataset, 5 tweets were used for manual implementation of the K-Means algorithm calculation, through a series of pre-processing processes, including TF-IDF weighting. After the manual K-Means calculation, from 29 words generated from TF-IDF, the following clustering results were obtained: Cluster 0 (positive) contains 5 words, Cluster 1 (neutral) contains 18 words, and Cluster 2 (negative) contains 6 words. These results show that the K-Means algorithm can effectively cluster sentiment in public opinion data related to the 2024 presidential election based on patterns found in the words in the tweets.