[Paper Review] On the relevance of data science for flight delay research: a systematic review

Alice Sternberg, Jorge de Abreu Soares+2 more|arXiv (Cornell University)|Dec 24, 2020

Air Traffic Management and Optimization211 references31 citations

TL;DR

This systematic review examines data science approaches to flight delay prediction, proposing a taxonomy based on scope, data, and computational methods—particularly emphasizing machine learning. It identifies trends, key works, and methodological evolution, offering a comprehensive framework for improving prediction accuracy in air traffic management.

ABSTRACT

Flight delays hurt airlines, airports, and passengers. Their prediction is crucial during the decision-making process for all players of commercial aviation. Moreover, the development of accurate prediction models for flight delays became cumbersome due to the complexity of air transportation system, the number of methods for prediction, and the deluge of flight data. In this context, this paper presents a thorough literature review of approaches used to build flight delay prediction models from the Data Science perspective. We propose a taxonomy and summarize the initiatives used to address the flight delay prediction problem, according to scope, data, and computational methods, giving particular attention to an increased usage of machine learning methods. Besides, we also present a timeline of significant works that depicts relationships between flight delay prediction problems and research trends to address them. The published version of this paper is made available at \url{this https URL}. Please cite as: L. Carvalho, A. Sternberg, L. Maia Goncalves, A. Beatriz Cruz, J.A. Soares, D. Brandao, D. Carvalho, e E. Ogasawara, 2020, On the relevance of data science for flight delay research: a systematic review, Transport Reviews

Motivation & Objective

• To analyze the role of data science in addressing flight delay prediction challenges in commercial aviation.
• To identify and categorize existing approaches based on scope, data sources, and computational methods.
• To highlight the growing adoption of machine learning techniques in flight delay modeling.
• To map significant research contributions and evolving trends over time through a chronological timeline.
• To provide a structured taxonomy to guide future research and system development in flight delay prediction.

Proposed method

• Conducting a systematic literature review of peer-reviewed studies on flight delay prediction from a data science perspective.
• Classifying existing approaches along three dimensions: research scope (e.g., airline-level, airport-level), data types (e.g., historical flight data, weather, air traffic), and computational methods (e.g., statistical models, machine learning, deep learning).
• Creating a taxonomy to organize and compare methodologies across studies, emphasizing methodological evolution.
• Mapping key publications chronologically to visualize research trends and technological shifts in flight delay modeling.
• Focusing on machine learning techniques such as random forests, gradient boosting, and neural networks as dominant approaches.
• Synthesizing findings to identify gaps, recurring challenges, and opportunities in data-driven flight delay prediction.

Experimental results

Research questions

• RQ1 What are the dominant data science methods used in flight delay prediction, and how have they evolved over time?
• RQ2 How do different data sources (e.g., flight schedules, weather, air traffic) contribute to prediction model performance?
• RQ3 What are the key methodological differences between traditional statistical models and modern machine learning approaches in this domain?
• RQ4 How do research trends in flight delay prediction correlate with advancements in data science and data availability?
• RQ5 What are the most significant challenges and limitations in current flight delay prediction models from a data science standpoint?

Key findings

• Machine learning methods, particularly ensemble models like XGBoost and LightGBM, have become the dominant approach in recent flight delay prediction research.
• The integration of diverse data sources—such as weather, airport congestion, and historical delays—significantly improves prediction accuracy compared to models using only flight schedules.
• A clear trend toward more complex, data-intensive models has emerged, reflecting the increasing availability of high-resolution flight and operational data.
• Despite methodological advances, challenges remain in model interpretability, generalization across different airports, and real-time deployment.
• The systematic review identifies a growing consensus on the importance of feature engineering and temporal modeling in enhancing prediction performance.
• The timeline of key works reveals a shift from rule-based and statistical models to sophisticated learning-based systems, especially post-2015.