Advanced Machine Learning Approaches for Accurate Flight Delay Prediction

Longhua Xu

2024

Abstract

Accurate forecasting of aircraft delays is imperative for minimizing financial losses and enhancing passenger satisfaction within the aviation industry. Precise delay predictions can substantially improve operational efficiency and foster greater passenger loyalty. This study investigates three advanced machine learning methodologies for flight delay forecasting using the Kaggle dataset: Neural Networks (NN), Wide\& Deep Learning, and Categorical Boosting (CatBoost). NN leverages deep learning architectures to identify complex patterns in the data. Wide\& Deep Learning is a classic model combining low-level and high-level features. CatBoost is a model for a gradient-boosting algorithm created specifically to manage category information. The conclusion is that NN achieves a 0.8103 accuracy rate, Wide&Deep achieves a 0.8117 accuracy rate, and CatBoost achieves a 0.8363 accuracy rate. This study shows that different machine-learning techniques are good for other types of samples. By meticulously comparing the performance of NN, Wide\& Deep Learning, and CatBoost, our research enhances aviation operational efficiency and highlights the significance of tailored algorithms for handling categorical data in complex prediction tasks.

Paper Citation

in Harvard Style

Xu L. (2024). Advanced Machine Learning Approaches for Accurate Flight Delay Prediction. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 261-266. DOI: 10.5220/0013515500004619

in Bibtex Style

@conference{daml24,
author={Longhua Xu},
title={Advanced Machine Learning Approaches for Accurate Flight Delay Prediction},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={261-266},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013515500004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Advanced Machine Learning Approaches for Accurate Flight Delay Prediction
SN - 978-989-758-754-2
AU - Xu L.
PY - 2024
SP - 261
EP - 266
DO - 10.5220/0013515500004619
PB - SciTePress