Back to results list
Please use this identifier to cite or link to this item:
|Title:||A decision rule-based forecasting model for tourism demand : an application and comparison||Authors:||Goh, Ka-leng Carey||Keywords:||Hong Kong Polytechnic University -- Dissertations
Tourism -- Forecasting
|Issue Date:||2004||Publisher:||The Hong Kong Polytechnic University||Abstract:||Although the major goal of forecasting is to estimate the likelihood of future state of phenomena, a forecasting model is developed on the basis of many past and current data related to the phenomena. A large volume of studies in various fields, including tourism has examined this information embodied in related factors. While existing research has proven that non-economic factors such as psychological, anthropological, and sociological factors play as important a role as their economic counterparts in influencing travel motivation and destination choice, the non-economic factors have not been sufficiently analyzed in the studies of tourism demand. Existing tourism demand forecasting literature has primarily relied on conventional econometric models which are founded on strict statistical assumptions and stringent economic theory. The statistical assumptions state that the values of demand variable are normally distributed, the error terms of demand variables are constant over time and they are not correlated, and that the explanatory variables are not linearly correlated to each other. Moreover, these econometric models have been analyzed based on conventional economic theory, i.e., utility theory and consumption behavioral theory. Both theories suggest that economic factors, such as income, price, substitute price, advertising are the primary influences of demand. Although there exists a large amount of tourism literature that studies how these non-economic factors could affect travel motivation, and how travel motivation affects destination choices, published articles have failed to show how non-economic factors could affect demand for tourism. Given the complex and interrelated nature of today's tourism demand framework, the theoretical and statistical assumptions on which the econometric models are based may not accurately represent the real situation. This project intends to fill the gap in the literature by incorporating the rough sets theory to provide a systematic framework for studying imprecision and vagueness in tourism data. Unlike econometric models which employ statistical inferential techniques, a rough sets algorithm is based on data mining techniques and therefore, imposes no assumptions on the knowledge discovery process. As tourism is a multidisciplinary industry, demand for tourism is possibly influenced by more than just economic factors. Therefore, the analysis of tourism demand should go beyond the orthodox demand framework to allow for more important factors, both numeric and non-numeric, economic and non-economic to be included in the analysis. Because rough sets algorithms can handle both numeric and non-numeric data, the non-economic factors, along with economic factors can be analyzed without having to bear the problem of insufficient degree of freedom and loss of information from transforming the non-numeric factors into numeric ones.
As agreed by many researchers, the capabilities of traditional relationship representation and forecasting models have long been known; they have been widely exploited and are close to their possible peak of applications. Only marginal improvements can be achieved. Many times, these models are expanded at the expense of model comprehensiveness. In such models, demand relationship is represented in a complicated system of equations, which policy makers and practitioners hardly comprehend. With the rough sets analysis, the relationship is depicted in a set of rules which are explicit and simple to understand. Moreover, a decision rule always consists of a single and relatively independent piece of information. In contrast to the system of related equations which the econometric model represents, the modular nature of decision rules make it possible for industrial practitioners and researchers to adapt and modify existing decision rules without affecting the overall system. To incorporate qualitative factors into the tourism demand analysis, this study constructs two socio-psychological attributes - leisure time and climate. Issues of data categorization for decision and condition attributes involving trending behavior typical in the context of time series forecasting are also addressed. To facilitate a comparison between rough sets model and econometric models, this study also investigates the methodological issues pertaining to two most popular econometric models found in the recent tourism literature. The chapter also proposes a method for accuracy performance comparison between econometric models and rough set models. Several important findings have been generated in this study. First, tourism data are relatively "rough" implying that probability distribution of tourism data might not follow the statistical assumptions required for many estimation procedures in econometric analysis. Second, as indicated by attribute strengths, climate and leisure appear to be more important than other economic factors in explaining changes in tourism demand. Third, the weight of each attribute in explaining changes in tourism demand has been found to be different at different levels of tourism demand. Most importantly, rough sets models outperformed the two econometric models in terms of forecast accuracy. This finding is encouraging as it confirms the ability of the rough sets theory in complementing the conventional methods in analyzing the forecasting tourism demand.
|Description:||xiii, 492 p. : ill. ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P HTM 2004 Goh
|URI:||http://hdl.handle.net/10397/2296||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b17726748_link.htm||For PolyU Users||179 B||HTML||View/Open|
|b17726748_ir.pdf||For All Users (Non-printable)||11.53 MB||Adobe PDF||View/Open|
Citations as of Dec 17, 2018
Citations as of Dec 17, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.