Sentiment analysis of the Indonesian community toward face-to-face learning during the Covid-19 pandemic
Andrew Giovanni Gozal, Hady Pranoto, Muhammad Fikri Hasani · Procedia Computer Science · 2023
This research aims to analyze how the opinions of the Indonesian people in the learning system in the middle of the covid-19 pandemic. The research method is carried out by performing the K-NN method to determine the accuracy level of the data used. The research data method is taken through public comments on Twitter social media using scrapping techniques with appropriate keywords. The data will then be processed through training and testing and classified using the K-NN method. After the data is classified, the accuracy, F1-Score, Recall, and Precision level will be tested using Confusion Matrix. The result showed that KNN performed well, with above 70% of the F1-score for each class. According to the confusion matrix, accuracy also showed promising results with 82%. Future research may include oversampling the class with fewer numbers. K-Fold cross-validation can also be used to see the general performance of the model. The same method may be used to find sentiment towards a political policy that is taken; whether a policy gets a good or bad response, if the response is bad, you can see the causal factors that cause the negative sentiment. In this way, you can find out the will of the public.