Study The SVM Kernel For Classification Of Covid-19 Vaccine Data On Twitter

Abstract

The rapid development of Covid-19 in Indonesia in 2020 caused the government too blige all Indonesian people to carry out the Covid-19 vaccination. The public's response to the policy, some agree and some disagree. The response is widely pouredon social media, one of whichis Twitter. Social media Twitter is ranked 5th in the category of the most used social media with a user percentage of 56%. This shows that there is a very large opportunity for data sources that can be used to find out positive and negative public sentiment regarding the Indonesian government's policy regarding Covid-19 vaccination. The method used to classify in this research is Support Vector Machine with various kernels. The kernels used are Linear, RBF, Polynomial, and Sigmoid. The classification results using the kernel are that the RBF kernel produces an accuracy of 88.8%, the Linear kernel produces an accuracy of 88.3%, the Sigmoid kernel produces an accuracy of 87% and the Polynomial kernel produces an accuracy of 85.5%. Based on the classification process that has been carried out, the highest accuracy is generated by the RBF kernel and the lowest accuracy is generated by the Polynomial kernel.