The chaos on US domestic airline passenger demand forecasting caused by COVID-19. (10th May 2022)

• Record Type: Journal Article
• Authors: Jafari, Nahid
• Is Part Of: International journal of business forecasting and marketing intelligence. Volume 7:Number 3(2022)
• Page Start: 241
• Page End: 258
• Publication Date: 2022-05-10
• Subjects: air passenger demand -- the US airlines market -- seasonal time series forecasting -- deep learning -- gated recurrent units; Business forecasting -- Periodicals; Marketing research -- Periodicals; 658.40355
• Languages: English
• ISSNs: 1744-6635

Abstract

Commercial aviation is a major contributor to the US economy, directly or indirectly generating approximately US$680 billion, or 4% of GDP, and supporting millions of jobs. Approximately 965 million passengers flew to US destinations in 2017 (https://rosap.ntl.bts.gov/view/dot/37861). Given the importance of the industry, accurate forecasting of air passenger demand is valuable, and the most sophisticated forecasting technologies can be applied to this endeavour. The ongoing COVID-19 crisis has had an unprecedented impact on air traffic. Effective forecast of passenger demand would benefit airlines to develop adequate recovery plans and prevent (or minimise) any catastrophe in handling passengers during and post pandemic. The purpose of this study is to investigate COVID-19's impact on the US domestic air passengers demand, identify the most influential features on air passenger demand, and design more accurate forecast models. In addition, we address a computational challenge in developing forecasting models due to the volatility of the recent data as a result of the COVID-19 crisis. We use both traditional and artificial intelligence methods and discuss their capabilities to handle the challenge.