Insights into TripAdvisor's online reviews: The case of Tehran's hotels. (April 2020)

• Record Type: Journal Article
• Authors: Khorsand, Ramina; Rafiee, Majid; Kayvanfar, Vahid
• Is Part Of: Tourism management perspectives. Volume 34(2020)
• Publication Date: 2020-04
• Subjects: Customer rating -- Data mining -- Hospitality -- Machine learning -- Online hotel reviews; Tourism -- Periodicals; 338.4791
• DOI: 10.1016/j.tmp.2020.100673
• Languages: English
• ISSNs: 2211-9736

Abstract

Abstract: User-generated data in TripAdvisor.com consists of considerable amount of useful information that can help managers to provide better services to their customers. This study aims to forecast a new user's rate to a hotel based on information of the hotel and user. To do so, all reviews on all hotels of Tehran on TripAdvisor.com as real data are selected and 8 different supervised machine learning models are applied to the data to select the best method including K-nearest neighbors (KNN), Naïve Bayes, decision tree, logistic regression, support vector machine, neural network, random forest, and gradient boosting. KNN algorithm which uses similarity and distance measures for classification is selected as the best method through conducted comprehensive comparisons, statistical analysis and data-based sensitivity analysis. Since this study investigates an intensive set of data of all hotels in a city in all time, some worthful managerial insights are presented. Highlights: Applying 8 machine learning models to predict the new users' rates in TripAdvisor. Conducting extensive comparisons over employed methods to find the best model. Implementing data-based sensitivity analysis to the best obtained model. Using the data of a real case study of all Tehran's hotels in TripAdvisor. Presenting worthful managerial insights for practitioners and managers of hotels.