Machine Learning based Social Media Sentiment Analysis for Predicting Flight Delays in US Airlines

Social Media is crucial to understanding customers' feelings, emotions, and thoughts about their experiences anywhere. Leveraging social media sentiment helps organizations improve strategy and decision-making for a company or organization will greatly help in making decisions that lead to the company’s success and continuity. In this study, we performed sentiment analysis on tweets related to air travel to analyze common emotions and their underlying causes. The findings indicate that negative sentiments are most frequently expressed, with flight delays being the primary cause of dissatisfaction. Using this insight, we built a predictive model to estimate flight delays based on sentiment and historical flight data to estimate the likelihood of flight delays using historical flight data and relevant features. We evaluated the model usingmultiple machine learning algorithms, achieving accuracy scores of 78.52% with Random Forest, 79.93% with XGBoost, and 37% with Logistic Regression. These results highlight the practical role of sentiment analysis in predictive modeling, demonstrating its potential to improve travel experiences by predicting flight delays and enabling airlines and passengers to make more effective proactive decisions.