Forecasting models for Portugal’s inbound tourism

• Autor(a) principal: Dias, Nuno Miguel de Castro de Amorim Oliveira
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/30720
• topic: Tourism forecasting; Portugal’s tourism; Forecasting models; Previsão do turismo; Turismo de Portugal; Modelos de previsão

In Portugal, the tourism sector has a considerable weight in the GDP and, as Small and Medium Enterprises (SMEs) were the most impacted by the pandemic, any contribution to help them to recover is welcome. It is in this context that the European RESETTING project has been setup, aiming to provide tools that can help these SMEs. This research work fits into the purpose, by studying and providing tourism forecasting models so Portuguese SMEs can use them and adjust their products and services to tourism demand. After analysing the research field related to tourism forecasting, we have figured out the most popular algorithms and what data should be used. This research study follows a well-known methodology to deal with data – CRISP-DM. Hence, we collected and prepared the data of interest, then we analysed it so we could better understand it. Finally, we have created two classes of forecasting models: baseline and deep learning forecasting models. The more complex deep learning models were based on Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) algorithms. From the experiments carried out we concluded that both deep learning models work well with data of daily frequency and were outperforming the baseline models. However, it did not work that well in the case of monthly data, when comparing with the baseline models. In the end, the models provided can help SMEs to obtain tourism demand predictions, both for the near-term – nowcasting – and the medium-term horizons.