Text Classification for Analysing Indonesian People's Opinion Sentiment for Covid-19 Vaccination

Abstract

The purpose of this study is to implement text mining for sentiment analysis of Indonesian public opinion on COVID-19 vaccination on Twitter social media using text classification techniques Support Vector Machine (SVM) and Random Forest. The research begins with crawling data from Twitter from September 2021 to October 2021; data cleansing; text translation into English; data preprocessing using NTLK performed with and without the lemmatization process; sentiment analysis using TextBlob; distribution of training and testing data with the Hold-Out method of 70:30 and 80:20; hyperparameter tuning with GridSearchCV; text classification with SVM and Random Forest; and testing the classification results by calculating Accuracy, Precision, Recall, F-Measure based on confusion matrix. The results show that text classification Random Forest consistently has a higher accuracy rate than SVM with the highest accuracy value of 90,59% and most of the sentiments indicate neutral to the COVID-19 vaccination program.