CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Uncovering Flight Delay Patterns by using Subgroup Discovery technique

Estudiants que han llegit aquest projecte:

Director/a: KULJANIN, JOVANA

Departament: FIS

Títol: Uncovering Flight Delay Patterns by using Subgroup Discovery technique

Data inici oferta: 24-01-2025     Data finalització oferta: 24-09-2025


Estudis d'assignació del projecte:
    GR ENG SIST AEROESP
Tipus: Individual
 
Lloc de realització: EETAC
 
Paraules clau:
Airline Delay, Cost, Airline Sustainability
 
Descripció del contingut i pla d'activitats:
This thesis investigates the phenomenon of flight delays in commercial aviation through the application of advanced data analysis and unsupervised machine learning techniques. Using a large anonymised operational dataset from a European airline, the study develops a multi-stage methodology encompassing data engineering, cleaning and preparation, exploratory statistics, and the application of Subgroup Discovery to uncover hidden patterns of delays.

The research first provides an overview of the operational dataset and the main variables influencing delay performance, such as temporal factors, city-pair concentration and flight duration. Exploratory analysis identifies recurrent trends and contextualises the distribution of delays across years. Subsequently, Subgroup Discovery is applied as an innovative technique to reveal interpretable rules that explain under which conditions delays are more likely to occur, highlighting the impact of operational contexts such as time of day, seasonality and specific route profiles.

The findings contribute to a better understanding of the complex drivers of flight delays, offering insights that go beyond aggregate statistics. They also demonstrate how rule-based unsupervised learning methods can support airlines in identifying critical operational scenarios and mitigating inefficiencies. Finally, the study outlines the implications of delay reduction in terms of cost savings, environmental benefits and improved passenger experience, aligning with broader sustainability and competitiveness goals in the aviation sector.
 
Overview (resum en anglès):
This thesis investigates the phenomenon of flight delays in commercial aviation through the application of advanced data analysis and unsupervised machine learning techniques. Using a large anonymized operational dataset from a European airline, the study develops a multi-stage methodology encompassing data engineering, cleaning and preparation, exploratory statistics, and the application of Subgroup Discovery to uncover hidden patterns of delays.

The research first provides an overview of the operational dataset and the main variables influencing delay performance, such as temporal factors, city-pair concentration and flight duration. Exploratory analysis identifies recurrent trends and contextualises the distribution of delays across years. Subsequently, Subgroup Discovery is applied as an innovative technique to reveal interpretable rules that explain under which conditions delays are more likely to occur, highlighting the impact of operational contexts such as time of day, seasonality and specific route profiles.

The findings contribute to a better understanding of the complex drivers of flight delays, offering insights that go beyond aggregate statistics. They also demonstrate how rule-based unsupervised learning methods can support airlines in identifying critical operational scenarios and mitigating inefficiencies. Finally, the study outlines the implications of delay reduction in terms of cost savings, environmental benefits and improved passenger experience, aligning with broader sustainability and competitiveness goals in the aviation sector.

© CBLTIC Campus del Baix Llobregat - UPC