Methods and Strategies for Air Travel Demand Forecasting in China Based on TEI@I Methodology

Abstract

Air travel demand forecasting plays a crucial important role in air transportation management. An accurate forecast of future travel demand is an indispensable basis of transportation planning, design and operations by both airport authorities and governments, and also an important determinant for increasing the profitability of airlines. The involving decisions may disperse widely according to time horizons, mainly including long-term government’s policy making and financial commitments, facilities expansion or planning, medium-term air-route planning, budgeting, evaluation of specific policies and short-term operations such as staffing, aircraft scheduling decisions, maintenance planning, advertising and sales campaigns and the opening of new sales offices etc.
Research about air travel demand forecasting methods and strategies is meaningful especially for developing countries like China, whose air transportation market is experiencing rapid growth. This industry now plays an increasingly important role in the development of national economics and represents an indispensable engine of economic growth. However, it’s a very risky industry to invest in because of the costly infrastructure expenses and perishable nature of the product it provides. Therefore, providing an in-depth study for Chinese air travel demand forecasting methods is of vital practical significance. And forecasting methods for various time horizons are required due to the widely disperse decisions.
Hence, in this thesis, we propose an integrated forecasting framework based on TEI@I methodology for Chinese air travel demand, and present studies about demand forecasting methods for short-term, medium-term and long-term, respectively. Main contributions of this thesis include:
1) Propose an integrated air travel demand forecasting framework based on TEI@I methodology, where the basic “decomposition, then integration” principle will guide us in next studies. This framework combines traditional mathematical forecasting models with experts’ knowledge, which plays an increasingly important role in demand forecasting especially for medium- and long-term future.
2) Provide a scientometric analysis for general demand forecasting research with CiteSpace software. With this computational tool, the whole intellectual landscape of demand forecasting, thematic patterns, landmark articles and emerging trends for future research are computationally detected and visually represented. This study has for the first time outlined the demand forecasting research’s evolutionary trajectory over the past decades and highlighted the emerging trends in future research, and in the meantime provide reference for research about air travel demand forecasting methods.
3) Propose an integrated forecasting framework with empirical mode decomposition method, to deal with the short-term forecasting problem particularly under complex and volatile economic circumstances. Under this framework, suitable linear and nonlinear models are developed for linear and nonlinear components, respectively, then all single forecasts are integrated to achieve the final forecast. One contribution of this study is that we propose a novel combination way to the combination forecasting family, to face high volatile circumstances.
4) Propose a short-term forecasting framework based on seasonal decomposition with a novel use of Google Trends data to quantify the moving holiday effect, especially to deal with the poor forecasting performance problem for those periods facing moving holiday effects. In the meantime, a nowcasting process is discussed and demonstrated through MIDAS regression model, with currently available weekly Google Trends data. One contribution of this study is that we present a novel way to apply Google Trends data in short-term air travel demand forecasting, which is a successful attempt and can be extended for other applications in further research.
5) Propose a demand forecasting method based on stochastic frontier analysis models and a model average technique to solve the unobservability problem of historical demand series, which has been always ignored. This is one of the first study focused on estimating and forecasting the underlying demand in demand forecastin literature, and an empirical application of this method for medium-term air travel demand forecasting is presented.
6) Propose an integrated long-term forecasting method for Chinese national air travel demand. Firstly, a cointegration analysis for air travel demand and its key determinants is implemented, and several scenarios are set for explanatory variables in trend extrapolation step. Secondly, we picture a possible long-term evolution pattern for Chinese air travel demand with other countries’ development experience. Integrating results from each functional module with experts’ knowledge, we achieve the final forecast for a long-term future, i.e., the year of 2020 and 2030. It is worth to note that in this integrated forecasting framework, experts’ knowledge plays a leading role in almost every step.

Date of Award	8 Jul 2016
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Kin Keung LAI (Supervisor)