Time Series Analysis in American Air Passengers

Tourism in too many areas has been increasing for decades because of development in communications, transportations, and increased awareness of opportunities through global economic integration. Recent events like Malaysia air crash or disappearance have negatively impacted Malaysian airline in many facets. One of the impacts is the decrease in the number of air passengers using Malaysian airline after those accidents. This emphasizes the importance of forecasting numbers of air passenger. There are two common approaches to forecasting international travel and tourism demand in the literature. Those are econometric models and time series models. This paper discusses different techniques in time series analysis, along with showing the relationship with economic cycles, natural disasters, wars, accidents, and man-made disasters in the United States air passenger numbers. Economic approach is often referred to as “structural demand modeling.” In terms of forecasting tourism demand, a problem with casual models is that values of the explanatory variables need to be predicted or known for the time at which the forecast is required. It is difficult to quantify some explanatory variables or proxy measures, which is sometimes biased (John Coshall, 2006). Lim (1997) decided not to employ economic demand modeling techniques. This paper therefore only shows some similarity between the economic factor and numbers for air traveling without trying to analyzing the relationship in further detail in terms of econometric model. Seasonality is an important feature of tourism demand time series and requires careful examination in modeling and forecasting seasonal tourism demand (Kulendran and Wong, 2005). Outlier and structural changes are commonly encountered in time series data analysis. The presence of those extraordinary events could easily mislead the conventional time series analysis prodedure resulting in erroneous conclusion. In research by Jennifer Christine, Hsin (2011) show that inbounds tourism from Japan has been severely affected by communicable disease. Man-made crisis and natural disasters have affected international tourism demand considerably (Song and Li, 2009). The effects of these events on tourism demand are to some extent predictable based on appropriate scenarios analysis (Song and Li, 2008). In the area of even tourism, neither time-series not structural equation approaches are of much use because special events are in a sense, statistical outliers (Tideswell, et.al. 2001). Therefore, regardless of different methods being used to forecast the future, the forecast would be totally off if there are special events happen. In this presentation, different forecasting methodologies will be explained and used. Including Moving Average, Exponential Smothering, Winters’ method, Decomposition, ARIMA, and Fourier Analysis. The forecasting accuracy comparison is conducted based on several measures of error magnitude: the mean absolute percentage error (MAPE), and the root mean square percentage error (RMSPE)