Sentiment Analysis of Presidential Candidates Anies Baswedan and Ganjar Pranowo Using Naïve Bayes Method

Abstract

Presidential elections in Indonesia are carried out in a democratic manner in which the people choose the figures who will nominate themselves for president. With the presidential nomination, there will be many surveys of several figures who have electability to become presidential candidates. Based on a survey that has been issued by several figures who are running for president, namely Anies Baswedan and Ganjar Pranowo, who are the benchmarks for the community to be able to express their opinions from existing social media, one of which is Facebook. This study takes data through a scraping process which is then cleaned or cleaned, then five labels are given, namely: 1 (very negative), 2 (negative), 3 (neutral), 4 (positive), and 5 (very positive). aims to see which sentiment is the highest given by warganet to the upcoming presidential election. This study concludes that netizens have negative sentiments towards figures in the upcoming presidential election. seen from the data randomly generated 49% negative comments, 35% positive comments and 16% neutral. In addition, from 510 data taken by classification using the Naïve Bayes method, as well as testing using the 10-fold cross validation method with Quadgram tokenization resulted in an accuracy of 42.75%, precision 42.10%, and recall 42.70%.