Continuous delay in flight is a major problem for the aviation community. It results in losses, schedule adjustments, and passengers dissatisfaction. While machine learning has been used to predict delays, most earlier research focuses only on accuracy and does not explain why delays happen. This study fills that gap by combining prediction models with explainable AI (XAI) at Hazrat Shahjalal International Airport (HSIA), Dhaka. We used flight and weather data from 2022 to create a complete dataset by cleaning data, creating new features, and combining information. We tested several models including KNN, Support Vector Machine, Decision Tree, Random Forest, AdaBoost, XGBoost and CatBoost. The results show that ensemble models did the best, with CatBoost and XGBoost reaching 95% accuracy. To make these models more understandable, we used the LIME method, which showed that weather and scheduling were the main factors influencing delays. This study uniquely applies explainable machine learning to predict flight delays in a developing country context, specifically at Hazrat Shahjalal International Airport in Bangladesh. Most previous studies focused on countries like the UK, USA, Saudi Arabia, or China etc. By providing accurate predictions along with clear explanations, the results show how airlines and airport authorities can improve scheduling, better use resources, and meet international delay standards. This can help enhance passenger satisfaction and strengthen airport resilience.
扫码关注我们
求助内容:
应助结果提醒方式:
