air passenger demand: holiday effects 7; long-term forecasting for 8, 26; medium-term forecasting for 7–8, 25–26; seasonal adjustment 7; short-term forecasting for 6–7, 25
air transportation: industry 1–6; role in China 5
air travel demand: of airport 19–20; demographic factors 22, 23; determinants of 21–25, 27; economic factors in 21–23; emergencies 22, 24–25; forecasting methods 19–21, 26–27; forecasting problems 6–8; geographic factors 22, 23; government policy 22, 24; integrated forecasting framework 45–46, 46; market structure 22, 24; moving holiday effect 95–96; nationwide 20–21; of O-D (origin-destination) pairs 19; social factors 22, 23–24
Akaike Information Criteria (AIC) 96, 99
ARDL (Autoregressive Distributed Lag) bounds testing approach: cointegration relationship 127–128, 130–133; model specifications 134; see also long-term forecasting
artificial intelligence (AI): short-term demand forecasting 6, 7; techniques 51–53
artificial neural networks (ANN) 13; ANN/back-propagation NN (BPNN) techniques 51–53; ANN-based nonlinear forecasting module 43; genetic programming 56; support vector machines 54–55
Australia 134, 141; historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
autoregressive conditional heteroscedasticity (ARCH) model, econometrical model 50–51
Autoregressive Integrated Moving Average (ARIMA) 6, 11–12, 18, 83; econometrical linear forecasting module 41–43, 48–49
autoregressive moving average (ARMA) model, forecasting comparison 118–119, 120, 120
back-propagation neural network (BPNN) techniques: artificial intelligence 51–53; forecasting comparison 118–119, 120, 120; forecasting process with 52
bases and bases management module, TEI@I methodology 43–45
Bayesian model averaging (BMA) 111
Boeing Company 22
calendar: factor 101, 106; Gregorian 95, 107; holiday 96, 100; moving holiday effect 95–96, 102, 108
Canada: historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
China: air passenger number vs GDP per capital 4; air passenger traffic and growth rate 3; air passenger traffic and logistic curve fitting 135, 136; air transportation markets 2–6; gross domestic product (GDP) 116, 117; growth of air transportation and economy 4; oil price 116, 117; parameters of logistic curve fitting 136; role of air transportation in 5; SiChuan earthquake 86; trade value 116, 117; urban population 116, 117
Chinese New Year (CNY) 94–96, 102, 107–108
CiteSpace 8–9; evolution of demand forecasting research 73–78; scientometric analysis for demand forecasting 66, 69, 70; software 79
Civil Aviation Administration of China (CAAC) 5, 116, 124, 131
combination forecasting process, TEI@I methodology 57–59, 61–62
Consumer Confidence Index (CCI) 20
demand forecasting methods 10–18; actual passenger traffic vs estimated demand 120; air travel 18–25; electricity demand 11–12; long-term 15–18; medium-term 13–15; multivariate methods 17; multivariate time series models 14–15; production function approach 15; research background 110–111; scenario-based methods 16; short-term 11–13; simulation-based methods 16–17; stochastic frontier analysis (SFA) model 111, 112–115; tourism demand 12–13; univariate statistical filters 14; univariate methods 17–18; see also air travel demand
demand forecasting research: average citations per year 72, 73; clusters of cited references 75; co-citation clusters in time series 75–77; co-citation network of references 74; co-occurrence network of disciplines for 70; disciplines involved in 70, 71; evolution of 73–78; hybrid intelligent forecasting system 76–77; intellectual structure of 74–75; keywords and references with surging citation 77–78; landmark articles in 78, 79; main keywords in 71–72; most cited articles in literature 72, 72; yearly citations in, on Web of Science 67
econometrical models: ARIMA/SARIMA models 48–49; autoregressive conditional heteroscedasticity (ARCH)/generalized ARCH (GARCH) model 50–51; common categories 47; TEI@I methodology 47–51; vector autoregressive (VAR)/vector error correction (VEC) models 49–50
electricity demand forecasting 11–12
empirical mode decomposition (EMD) method, air travel demand 84–85, 85
error correction models (ECM) 13
expert knowledge and judgment: demand forecasting 59–61; methods of incorporating 59; modeling process 60
forecasting models: artificial intelligence techniques 51–56; combination methods 57–59, 61–62; econometrical models 47–51; expert knowledge and judgmental adjustment 59–61; linear combination approach 57–58; nonlinear combination approach 58–59; TEI@I methodology 46–61; see also demand forecasting methods
forecasting problems: air travel 6–8; long-term 8; medium-term 7–8; short-term 6–7
France: historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
frequentist model averaging (FMA) 111
Frisch, Ragnar 47
Garfield, Eugene 68
GDP (gross domestic product) growth 128–129; assumptions for 138; long-term forecasting 132–133; variable 131
generalized autoregressive conditional heteroscedasticity (GARCH) model, econometrical model 50–51
genetic algorithms (GA) 13, 56
genetic programming (GP), artificial intelligence 56
Germany 134, 141; historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
Google Trends 7, 9, 94, 144; Genhol vs 96–99, 99, 102, 103; nowcasting 104–108; proposed method with 102, 103
Harvey, Andrew 95
holidays: Chinese New Year (CNY) 94–96, 102, 107–108; Easter 99, 106, 107; moving holiday effect 95–96, 102, 108; Thanksgiving holiday 107
Holt-Winters forecasting methods 12, 21, 27, 125
Hong Kong International Airport (HKIA) 82–83, 85–86, 92, 144; historical air passenger traffic 98, 107–108; monthly passenger traffic of 86
Improved Particle Swarm Optimization (IPSO) algorithm 17
integrated short-term forecasting framework: adoption of SARIMA models 84–85; data description and evaluation criteria 86–87; empirical analysis 85–91; empirical mode decomposition (EMD) method in 84–85, 89, 91; general framework for short-term air travel demand 85; model comparisons 89, 90, 91, 91; modeling the linear component 87–88, 89; modeling the nonlinear component 88–89; proposed framework 83–85; TEI@I methodology basis in 82–83, 91–92
Intergovernmental Panel on Climate Change 130
International Air Transport Association (IATA) 1–2, 125, 138
International Energy Agency 130
Japan 134, 141; historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
Kalman-filter algorithm 15
Lagrange’s theorem 58
least squares support vector regression (LSSVR) 20, 54–55, 100, 101; demonstration of 55; nonlinear modeling 84–85, 88–89; seasonal decomposition (SD)_LSSVR model 101, 102, 103
long-term forecasting: air travel demand 8; ARDL bounds testing approach 132–133, 140–141; ARDL model specifications 134; cointegration relationship 126, 127–128, 130–133; economic globalization and urbanization 124–126; empirical analysis 130–138; empirical data 130, 131; fuzzy linear regression equations 18; general framework 126; irregular event effects 127, 129, 135–137; logistic growth curve fitting 133–134, 135, 136; long-term evolution pattern 126–127, 128–129; Markov-switching regime 129, 135–137; methods 15–18; multivariate methods 17; proposed framework for 126–130; scenario-based methods 16; scenario planning 130, 137–138, 139, 140; simulation-based methods 16–17; unit root testing 131–132, 133; univariate methods 17–18
MAPE (Mean Absolute Percentage Error) 87, 98; equation 87; forecasting performance 91, 99, 102, 103, 106, 119–120, 120, 131, 132, 141; forecasting vs nowcasting 106
Markov-switching (MS) regime approach 129, 135–137, 141
medium-term forecasting: actual passenger traffic vs estimated demand 120; air travel demand 7–8; methods 13–15; multivariate time series models 14–15; production function approach (PFA) 15; univariate statistical filters 14; see also stochastic frontier analysis (SFA) model
MIDAS (mixed data sampling) model 96; MATLAB toolbox 106; methodology of 104–105; nowcasting results 105–106, 107
model average (MA): comparison of stochastic frontier analysis (SFA) model and 118–121; appendices for calculations 147–150
Monte Carlo simulation, long-term forecasting 16–17
multivariate methods: long-term forecasting 17; time series models for medium-term forecasting 14–15
nationwide air travel demand: long-term forecasting for 8; medium-term forecasting for 7–8; short-term forecasting for 6–7
Natural Bureau of Statistics of China 3, 131
nonlinear autoregressive (NAR) model 53
nowcasting: forecasting vs 106; Google Trends 104–108; MIDAS (mixed data sampling) model 104–105; MIDAS with leads 105; results 105–106, 107
Quantum-Behaved Particle Swarm Optimization (QPSO) algorithm 17
research see demand forecasting research
RMSE (Root Mean Squared Error) 87, 98, 102; equation 87; forecasting performance 91, 102, 103, 106, 119–120, 120
scenario-based methods, long-term forecasting 16
Science Citation Index 68
Science Citation Index Expanded (SCI-EXPANDED) 70
scientometric analysis 66–69; bibliographic records collection 69–70; co-citation clusters in time series 75–77; disciplines involved in demand forecasting 70, 71; keywords and references with surging citation 77–78; main keywords in demand forecasting 71–72; most cited articles in demand forecasting literature 72, 72; overview of bibliography 69–72, 73; see also demand forecasting research
Seasonal Autoregressive Integrated Moving Average (SARIMA) 6, 11, 13, 83, 95, 101; econometrical forecasting model 48–49; forecasting linear component 84, 87–88
seasonal decomposition (SD) forecasting: empirical analysis 101–103; Genhol vs Google Trends 96–99; hybrid forecasting method 99–103; moving holiday effect 95–96; overall forecasting process 101; proposed framework 100; X-13-ARIMA-SEATS method 96–97, 100, 101
seasonality 2, 7, 11–12, 47; air travel demand and 87, 94–96; definition 94; historical passenger traffic at HKIA 86; moving holiday effect 95–96, 107
short-term forecasting: air travel demand 6–7; electricity demand 11–12; methods 11–13; nowcasting 104–106, 107–108; seasonality and 94–96; tourism demand 12–13; see also integrated short-term forecasting framework
Shuffled Frog-Leaping (SFL) algorithm 17
SiChuan earthquake 86
simulation-based methods, long-term forecasting 16–17
Social Sciences Citation Index (SSCI) 70
STAMP method 95
statistical filters: Hodrick-Prescott filter 7, 14; univariate 14, 25
statistical learning theory (SLT), support vector machines (SVM) 54
stochastic frontier analysis (SFA) model: appendices for calculations 147–150; application to air travel demand forecasting 116–121; data for 116–118; demand estimation 115; demand forecasting and 111, 115; evaluation and comparison to model average (MA) 118–121; methodology 112–115; model average 112–114; model construction 115; proposed demand forecasting framework 115; raw data collection 114–115; summary of variables 116
support vector machines (SVM) 13
TEI@I methodology 8, 34–35, 61–62, 124, 143; ANN-based nonlinear forecasting module 43; ARIMA-based econometrical linear forecasting module 41–43; back-propagation neural network (BPNN) and forecasting process 43, 44; bases and bases management module 43–45; common forecasting models 46–61; general framework of 35–45, 36; integrated air travel demand forecasting framework 45–46; integrated short-term forecasting 82–83, 91–92; man-machine interface (MMI) module 36; rule-based expert system module 40–41; web-based text mining (WTM) module 37, 37–40; see also integrated short-term forecasting framework
United Nations 130
United States 134, 141; historical air passenger traffic and logistic curve fitting 135; logistic curve fitting 133; parameters of logistic curve fitting 136
univariate methods: long-term forecasting 17–18; medium-term forecasting 14; statistical filters 7, 14
US Department of Defense 130
vector autoregressive (VAR) models 13; econometrical model 49–50
vector error correction (VEC) model 127; econometrical model 49–50
web-based text mining (WTM) module: component of TEI@I methodology 37–40; feature extraction phase 37, 38; main processes of 37; structure analyzing phase 37, 38; text classification phase 37, 39; see also TEI@I methodology
Web of Science database 8, 66, 69–70; yearly citation in demand forecasting 67; yearly published articles about demand forecasting 67
Web of Science™ Core Collection 70, 79
World War II 130