A Holistic Approach on Airfare Price Prediction Using Machine Learning Techniques

Abstract

Globalization of markets involves new strategies and price policies from professionals that contribute to global competitiveness. Airline companies are changing tickets’ prices very often considering a variety of factors based on their proprietary rules and algorithms that are searching for the most suitable price policy. Recently, Artificial Intelligence (AI) models are exploited for the latter task, due to their compactness, fast adaptability, and many potentials in data generalization. This paper represents an analysis of airfare price prediction towards finding similarities in the pricing policies of different Airline companies by using AI Techniques. More specifically, a set of effective features is extracted from 136.917 data flights of Aegean, Turkish, Austrian and Lufthansa Airlines for six popular international destinations. The extracted set of features is then used to conduct a holistic analysis from the perspective of the end user who seeks the most affordable ticket cost, considering a destination-based evaluation including all airlines, and an airline-based evaluation including all destinations. For the latter cause, AI models from three different domains and a total of 16 model architectures are considered to resolve the airfare price prediction problem: Machine Learning (ML) with eight state-of-the-art models, Deep Learning (DL) with six CNN models and Quantum Machine Learning (QML) with two models. Experimental results reveal that at least three models from each domain, ML, DL, and QML, are able to achieve accuracies between 89% and 99% in this regression problem, for different international destinations and airline companies.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 20 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.