چکیده مقاله
Air transportation continues to be one of the most popular modes of travel, primarily due to its ability to reduce travel time significantly However, flight delays partially erode this advantage, leading to substantial annual economic losses for both passengers and airlines This study focuses on flights at Kerman's Hashemi Rafsanjani International Airport and aims not only to predict delays but also to uncover the operational and environmental factors influencing their occurrence A comprehensive dataset was compiled from 2018 operational records and enriched with high resolution meteorological data temperature, wind speed, visibility, cloud cover, pressure, and precipitation Additional engineered features included aircraft pre delay, turnaround time, flight gap intervals, load factor, distance, and hour of day indicators to capture temporal and cumulative effects Multiple boosting classifiers CatBoost, XGBoost, LightGBM were trained to classify flights into three delay categories on time, short, and long LightGBM achieved the most balanced performance, with 94% overall accuracy and high precision recall across all classes SHAP analysis revealed that pre delay, turnaround time, and peak hour departures were the most influential predictors, followed by weather related variables These findings offer actionable insights for airport managers to optimize scheduling, mitigate cascading delays, and develop weather related contingency strategies
کلیدواژهها
نویسندگان
شیوه ارجاع
Khodadad, Mohammadhossein and Afshari Moez, Mohammad Amin and Samadzad, Mahdi,1404,Flight Delay Analysis Using Machine Learning: A Case Study of Kerman Airport,14th International Congress on Civil Engineering,Tehran
ارائهشده در
مجموعه مقالات چهاردهمین کنگره بین المللی مهندسی عمران29 مهر 1404 · تهران