Enhancing Airline Booking Prediction and Customer Sentiment Analysis Using Deep Learning and Multilingual Reviews

Shukla, Swati (2025) Enhancing Airline Booking Prediction and Customer Sentiment Analysis Using Deep Learning and Multilingual Reviews. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (332kB) | Preview

Abstract

In the growing world of data, the increasing availability of airline customer data presents new opportunities for improving booking prediction accuracy using machine learning and deep learning techniques to understand intricate patterns. This research investigates the impact of integrating multilingual sentiment analysis and demographic personalization on the performance of deep learning models for predicting airline bookings. Four publicly available datasets from Kaggle — Skytrax airline reviews, Twitter airline sentiment, Expedia booking data, and passenger satisfaction data — were merged to form a comprehensive multi-source dataset covering user reviews, booking outcomes, and customer attributes.

Sentiment scores were extracted using the XLM-RoBERTa transformer model to enable effective multilingual sentiment representation. Demographic features included age band, gender, travel class, and airline preferences. These features were embedded alongside sentiment vectors to personalize the prediction task. A range of models was developed, including traditional models such as Logistic Regression, Random Forest, Support Vector Machine, and KNN for baseline comparison, as well as a series of deep learning architectures progressing from text-only to hybrid models with fine-tuning and class weighting. Due to significant class imbalance, model performance was evaluated using ROC-AUC, F1 @0.30, and confusion matrices.

The results show that integrating sentiment and demographic features leads to consistent performance gains, with the Hybrid + Fine-Tuned model achieving the best performance (ROC-AUC = 0.6636, F1 @0.30 = 0.559). This research highlights the effectiveness of personalized, multilingual deep learning models in enhancing booking prediction accuracy for diverse airline customers and offers practical implications for CRM strategies in the airline industry.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Jameel Syed, Muslim UNSPECIFIED
Subjects:	H Social Sciences > HD Industries. Land use. Labor > Specific Industries > Aviation Industry H Social Sciences > HF Commerce > Marketing > Consumer Behaviour Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4150 Computer Network Resources > The Internet > World Wide Web > Websites > Online social networks T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > The Internet > World Wide Web > Websites > Online social networks
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	26 Aug 2026 12:04
Last Modified:	26 Aug 2026 12:04
URI:	https://norma.ncirl.ie/id/eprint/9668

Actions (login required)

View Item