Modelação da procura turística: um estudo comparativo entre redes neuronais artificiais e a metodologia de Box-Jenkins Artigo de Conferência uri icon

resumo

  • The present research aims to explore and to evidence the utility of the methodology of Artificial Neural Networks (ANN) in the analysis of tourism demand as an alternative to the Box-Jenkins methodology. The first methodology has arising interest in the economic and business area since several researches have verified that methodology presents a valid alternative to classical methods of forecasting allowing giving answer to situations in which the traditional ones will be of difficult to apply (Thawornwong & Enke, 2004). According to Hill et al. (1996) and Hansen et al. (1999) ANN show capacity to improve the time-series forecasts through of additional information analysis decreasing their dimension and reducing their complexity. For that, each one of the referred methodologies focused in the treatment, analysis and modeling of the tourism time-series: Monthly Guest Nights in Hotels registered between January 1987 to December 2006, since it is one of the variables that better explain the effective tourism demand. The Study was performed for two regions of Portugal: North region and Centre region. Considering the results, and according to the Criteria of MAPE for model evaluation proposed by Lewis (1982), the ANN model presented acceptable statistical qualities and adjustments satisfied. Being so, it is adequate not only for the modelling but also to the prediction of times series, when compared to the model performed by Box- Jenkins methodology. We intended also to evaluate the performance and competiveness of the tourism destinations - North region and Center region of Portugal - by main origin markets and to analyse how it is distributed their portfolio of origin markets for the period of 1997 to 2006. The Market Share Analysis tool proposed by Faulkner (1997) was applied and it was observed an high dependency of the domestic market for both regions.

data de publicação

  • janeiro 1, 2008