Scholarlink

<p>Aviation plays a significant role in global economic activities. However, flight delays, whether caused by weather or technical problems, remain a persistent issue, especially in areas exposed to extreme climates. This study examines the impact of meteorological features on departure flight delays at Kuwait International Airport, using data from 2012 to 2019. The main question of research is whether weather-related features, combined with operational variables, can enhance the performance of machine learning (ML) techniques in predicting flight delays. Four classification algorithms, including Logistic Regression, Decision Tree, Neural Network, and Ensemble Learning, were applied to a dataset that combined standard flight information with engineered features such as lagged weather conditions, categorized temperature regimes, and delay propagation indicators. Feature selection was performed using the Minimum Redundancy Maximum Relevance (MRMR) method to emphasize the relevance and reduce redundancy. Results show that dust storms and precipitation were the most influential variables impacting delays. All models achieved comparable predictive performance, with the Ensemble model performing best (Accuracy = 70.5%, TPR = 82.6%, F1-score = 77.3%). Therefore, the study that involving weather-driven and operationally lagged factors introduces practical enhancements to predict flight delay in extreme climates. The findings of this study contribute to the current studies by revealing that machine learning models can be applied to predict flight delay in Kuwait’s unusual environment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting departure flight delays due to weather at desertic airports using machine learning

  • 期刊论文

  • 发表日期：

    2026年2月12日

查看全文
  • Sharaf AlKheder,
  • Ashjan Alotaibi,
  • Reem Abdulrahman

摘要

Aviation plays a significant role in global economic activities. However, flight delays, whether caused by weather or technical problems, remain a persistent issue, especially in areas exposed to extreme climates. This study examines the impact of meteorological features on departure flight delays at Kuwait International Airport, using data from 2012 to 2019. The main question of research is whether weather-related features, combined with operational variables, can enhance the performance of machine learning (ML) techniques in predicting flight delays. Four classification algorithms, including Logistic Regression, Decision Tree, Neural Network, and Ensemble Learning, were applied to a dataset that combined standard flight information with engineered features such as lagged weather conditions, categorized temperature regimes, and delay propagation indicators. Feature selection was performed using the Minimum Redundancy Maximum Relevance (MRMR) method to emphasize the relevance and reduce redundancy. Results show that dust storms and precipitation were the most influential variables impacting delays. All models achieved comparable predictive performance, with the Ensemble model performing best (Accuracy = 70.5%, TPR = 82.6%, F1-score = 77.3%). Therefore, the study that involving weather-driven and operationally lagged factors introduces practical enhancements to predict flight delay in extreme climates. The findings of this study contribute to the current studies by revealing that machine learning models can be applied to predict flight delay in Kuwait’s unusual environment.