From Data to Strategy: Explainable AI Insights for Enhancing Airline Passenger Satisfaction and Loyalty

Abstract

Passenger satisfaction has become a cornerstone of competitiveness in the airline industry, directly influencing customer loyalty and long-term profitability. This study leverages artificial intelligence to both predict satisfaction outcomes and uncover the service factors that most strongly shape passenger perceptions. Using a large-scale real-world dataset, this study evaluates a comprehensive set of machine learning models. Among them, LightGBM has achieved the best performance, reaching 96.2% accuracy, 95.6% F1-score, and AUC 99.4%, establishing it as the most reliable predictive model. To ensure robust and actionable insights, feature importance was analyzed using three complementary techniques: permutation importance, gain-based measures, and Shapley Additive exPlanations (SHAP) values. Across all methods, in-flight Wi-Fi service and online boarding consistently emerged as the most decisive determinants of satisfaction, while seat comfort, cleanliness, baggage handling, and in-flight entertainment provided additional, yet meaningful, contributions. Together, these findings highlight that digital connectivity and operational efficiency significantly impact the passenger experience, complemented by traditional service dimensions. For airlines, the results translate into a clear strategic directive: prioritizing investment in high-quality Wi-Fi, seamless boarding, and continuous improvement in comfort and service standards will not only elevate passenger satisfaction but also strengthen loyalty, reduce churn, and ultimately maximize profitability.