A Hybrid Ensemble Machine Learning Approach For Arrival Flight Delay Classification Prediction Using Voting Aggregation Technique

Abstract

The number of flights keeps increasing with the development of civil aviation, and flight delays have become a severe issue of concern to the aviation industry. The need for reliable flight delay classification and prediction cannot be overemphasised because of its importance. In this research, we classify arrival flight delay using a hybrid ensemble machine learning algorithm based on voting aggregation. Each ensemble’s architecture consists of four supervised machine learning algorithms: Logistic Regression, Decision Tree, Support Vector Classifier and Random Forest. We conducted a comparative experiment using the United States Bureau of Statistics dataset to verify the efficacy of our proposed approach against other benchmark approaches. We employ multiple evaluation metrics to check the performance of our proposed model comprehensively. Accuracy, Recall, Precision, F1-Score, AUC Score, and ROC curve. Our experimental results show that our model outperforms the benchmark techniques on the binary classification of flight delays task. Hence, the hybrid ensemble model presented in this research can be used as a decision-support model to improve the flight scheduling management system design to assist the airline and airport operation managers and passengers with improved travel itineraries management.