Analysis of Public Sentiment of Covid-19 Dynamics on Social Media Using Support Vector Machine and Particle Swarm Optimization: Analisis Sentimen Publik Dinamika Covid-19 di Media Sosial Menggunakan Support Vector Machine dan Particle Swarm Optimization

Fakhri Muhammad (Universitas Singaperbangsa Karawang)
Chaerur Rozikin (Universitas Singaperbangsa Karawang)
Riza Ibnu Adam (Universitas Singaperbangsa Karawang)

Systematics
Vol 4 No 2 (2022): August 2022

Variants of covid in Indonesia continue to grow and make people required to stay at home and are not required to go out if they don't have important things, therefore many people who stay at home often play social media such as twitter, it is possible that many irresponsible people make opinions or hoaxes with a specific purpose to make tweets that are not in accordance with the facts, which are feared to make the public more panicked about the increase in this covid-19 variant. Therefore this study was conducted to classify tweets as positive, negative, and neutral. The methodology used is a text mining process with 4 modelings using Support Vector Machine and Particle Swarm Optimization. The results obtained from the 4th modeling produce an accuracy of 83% on the linear kernel. While the PSO modeling in scenario 4 with 90:10 data division resulted in the highest accuracy in linear and polynomial kernels of 86% and 87%, respectively. Other evaluation values ​​also improved, such as precision to 90%, recall to 83%, and f1-score to 86%.