Sentiment Analysis of Air Quality to Identify Public Perspectives in Indonesia Based on Machine Learning

The worsening air quality in Indonesia, especially in DKI Jakarta, has become a serious concern. In this context, sentiment analysis can provide new information about people’s views and responses to this environmental issue. In this study, we collected data from social media, i.e., the YouTube platform, which reflects people’s views on air quality. We implemented various ML algorithms such as Random Forest, Decision Tree, Naive Bayes, Extra Trees, and Hard Voting Ensemble to analyze the sentiment in the retrieved YouTube comments. We evaluate the performance of each algorithm based on the accuracy in classifying the sentiments. Through sentiment analysis, we identified the most common patterns of views appearing in the analyzed data. The novelty of the research is the application of Machine Learning techniques in analyzing sentiments related to air quality in Indonesia. As a result, all ML algorithms showed accuracy above 90%, except the NB algorithm with 76.14% accuracy. This shows that ML algorithms have good sensitivity to the dataset containing YouTube comments used in this study.