Analisis Sentimen Program Jaminan Kesehatan Nasional Menggunakan Multiclass Support Vector Machine

Abstract

Optimizing the implementation on National Helath Insurance which requires the use of BPJS participant cards in various public services is one of the government policies that is widely discussed and has garnered many opinions in the community. Public opinion is expressed through social media, one of which is through Twitter. The aim of this research is to classify public opinion regarding the new regulations of the National Health Insurance Program as a form of government policy to implement Presidential Instruction Number 1 of 2022 using Twitter data. Public opinion as many as 1.179 tweets were labeled positive, negative and neutral sentiments, then TD-IDF wighting was carried out and analyzed using the multiclass SVM algorithm with the One Against All approach. The results of the analysis showed that Multiclass SVM with a linear kernel was able to classify with an accuracy level of 81% where the classification pf positive sentiment was17 (7.6%), negative sentiment was 115 (48.7%) and neutral sentiment are 104 (44.1%). This shows that public sentiment is dominated by negative sentiment or disagreement with the new regulations of the National Health Insurance Program.