| This research focuses on the methodological means of providing recreation and tourism demand forecasts. It examines different statistical techniques, applying them to data on two case studies. The overall purpose of this study is to assess various means of demand forecasting. The first case analysis in this study uses three selected U.S. national park data sets, including the Great Smoky Mountains National Park, the Yellowstone National Park, and the Yosemite National Park. The second case study uses the selected U.S. zoological park (Milwaukee County Zoo, Wisconsin) monthly data sets. The national park data sets refer to annual visitation figures and were supplied by the National Park Service. In the case of zoological park visitors, the data sets were obtained from the Milwaukee County Zoo, Wisconsin.;This study employed the MAPE and RMSPE measuring forecast accuracy approaches. The MAPE values have indicated that the ARIMA method is more accurate than other approaches to predict the future visitation figures in both annual and seasonal data forms. From the MAPE and RMSPE values examined, those of ARIMA, Naive 1, and double exponential smoothing methods are, in the case of three national parks, superior to those of Naive 2, single exponential smoothing, and Holt's methods. ARIMA and SMA had the lowest RMSPE, and DES had the highest RMSPE among all the forecasting methods for the Milwaukee County Zoo. ARIMA outperformed all the other techniques in forecasting attendance figures for the next 12-month prediction. DES has the highest MAPE and RMSPE values and consistently performed worst among all the other techniques in forecasting the number of zoo visitors.;Choice and evaluation of forecasting methods, discussion of the proposed forecasting methods and the proposed explanatory variables, special events and dummy variables in the forecasting environment, and the advantages/disadvantages among different forecasting methods are also presented. |
