Both data patterns and forecasting accuracy have been identified in the literature as important constructs in tourism demand forecasting. However, the relationship between the two was either ignored or addressed only indirectly. This study develops an innovative methodology to identify and categorize typical tourism demand patterns, and assesses the impact of the identified patterns on the accuracy of various tourism forecasting methods using a large data set. Four distinct patterns were identified: (1) a stable linear trend, (2) a nonlinear trend, (3) a wave-shaped trend, and (4) an abrupt change pattern. These four arrival data patterns affect the accuracy of forecasting models in a specific and systematic manner. Within the tourism industry, it might be feasible to categorize data based on a small number of typical, easily observable features. The relationships between the performance of the forecasting models and these data patterns could be exploited for optimal selection of a forecasting method, improving the accuracy of the forecasts.
The aim of Tourism Analysis is to promote a forum for practitioners and academicians in the fields of Leisure, Recreation, Tourism, and Hospitality (LRTH). As a interdisciplinary journal, it is an appropriate outlet for articles, research notes, and computer software packages designed to be of interest, concern, and of applied value to its audience of professionals, scholars, and students of LRTH programs the world over.