Open AccessOptimising the Particle Swam Optimazion Usage for Predicting Indonesia Presidential Election Result Period 2019-2024
Author(s) -
Dinar Ajeng Kristiyanti,
Normah Normah
Publication year - 2019
Publication title -
sinkron
Language(s) - English
Resource type - Journals
eISSN - 2541-2019
pISSN - 2541-044X
DOI - 10.33395/sinkron.v4i1.10149
Subject(s) - naive bayes classifier , support vector machine , presidential election , presidential system , politics , general election , political science , social media , bayes' theorem , public opinion , classifier (uml) , computer science , artificial intelligence , law , bayesian probability
Indonesia is a Democrat nation. A general election known as the PEMILU has become a tradition of the nation that is synonymous with political issues and leadership turnover. Social media is one place in expressing the opinions and aspirations of people including politics, Twitter is one of the social media used as a place for politicians including two couples of presidential candidate and vice president of INDONESIA in Campaign to win a vote in the elections of 17 April 2019. This research analyzes public opinion i.e. comments on Twitter accounts @jokowi, @KyaiMarufAmin, @prabowo, @sandiuno into two categories of positive and negative opinions by comparing the text classifier model Naïve Bayes and SVM, and the implementation of The PSO algorithm to obtain optimal accuracy results. The results of the study show Prabowo Sandi won the prediction of presidential candidate with the best accuracy result of 77.00% acquired model Naïve Bayes + PSO, and 86.20% acquired model SVM + PSO, with an increase in accuracy 7.5% on model SVM, And 2.1% on the model Naïve Bayes when compared before done optimization with PSO algorithm.