Leveraging national tourist offices through data analytics

Sérgio Moro, Paulo Rita, Cristina Oliveira, Fernando Batista, Ricardo Ribeiro

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)


Purpose: This study aims to propose a data-driven approach, based on open-source tools, that makes it possible to understand customer satisfaction of the accommodation offer of a whole country. Design/methodology/approach: The method starts by extracting information from all hotels of Portugal available at TripAdvisor through Web scraping. Then, a support vector machine is adopted for modeling the TripAdvisor score, which is considered a proxy of customer satisfaction. Finally, knowledge extraction from the model is achieved using sensitivity analysis to unveil the influence of features on the score. Findings: The model of the TripAdvisor score achieved a mean absolute percentage error of around 5 per cent, proving the value of modeling the extracted data. The number of rooms of the unit and the minimum price are the two most relevant features, showing that customers appreciate smaller and more expensive units, whereas the location of the hotel does not hold significant relevance. Originality/value: National tourist offices can use the proposed approach to understand what drives tourists’ satisfaction, helping to shape a country’s strategy. For example, licensing new hotels may take into account the unit size and other characteristics that make it more attractive to tourists. Furthermore, the procedure can be replicated at any time and in any country, making it a valuable tool for data-driven decision support on a national scale.

Original languageEnglish
Pages (from-to)420-426
Number of pages7
JournalInternational Journal of Culture, Tourism, and Hospitality Research
Issue number4
Publication statusPublished - 1 Oct 2018


  • Data analytics
  • Data mining
  • National tourist offices
  • Online reviews
  • Sensitivity analysis
  • Web scraping


Dive into the research topics of 'Leveraging national tourist offices through data analytics'. Together they form a unique fingerprint.

Cite this