Using machine learning and big data for efficient forecasting of hotel booking cancellations

Cancellations are a key aspect of hotel revenue management because of their impact on room reservation systems. In fact, very little is known about the reasons that lead customers to cancel, or how it can be avoided. The aim of this paper is to propose a means of enabling the forecasting of hotel booking cancellations using only 13 independent variables, a reduced number in comparison with related research in the area, which in addition coincide with those that are most often requested by customers when they place a reservation. For this matter, machine-learning techniques, among other artificial neural networks optimised with genetic algorithms were applied achieving a cancellation rate of up to 98%. The proposed methodology allows us not only to know about cancellation rates, but also to identify which customer is likely to cancel. This approach would mean organisations could strengthen their action protocols regarding tourist arrivals.     

Impact of financial/economic crisis on demand for hotel rooms in Hong Kong

The main objectives of this study are (1) to identify the factors that influence the demand for hotel rooms in Hong Kong and (2) to generate quarterly forecasts of that demand to assess the impact of the ongoing financial/economic crisis. The demand for four types of hotel room from the residents of nine major origin countries is considered, and forecasts are generated from the first quarter of 2009 to the fourth quarter of 2015. Econometric approaches are employed to calculate the demand elasticities and their corresponding confidence intervals, which are then used to generate interval demand predictions. The empirical results reveal that the most important factors in determining the demand for hotel rooms in Hong Kong are the economic conditions (measured by income level) in the origin markets, the price of the hotel rooms and the ‘word of mouth’ effect. Demand for High Tariff A and Medium Tariff hotel rooms is estimated to have experienced negative annual growth in 2009 due to the influence of the financial/economic crisis, whereas that for High Tariff B hotel rooms is thought to have grown in 2009 after having decreased in 2008. The demand for tourist guesthouse rooms is expected to be the least affected by the crisis. Overall demand is predicted to recover gradually from 2010 onwards.     

Identifying unreliable online hospitality reviews with biased user-given ratings: A deep learning forecasting approach

This study considers the review reliability problem by identifying biased user-given ratings through rating prediction on the basis of the textual content. Deep learning approaches were introduced to investigate the textual review and validate the effect of rating prediction using a dataset collected from Yelp. The definition of “biased rating” was clarified and influenced the matching rules. The approach obtains high performance on a total of 1,000,000 reviews for prediction, with user-given ratings as the benchmark. Using the revealed biased ratings, unreliable reviews were detected by combining the results of several deep learning kernels. Findings shed light on understanding review quality by distinguishing biased ratings and unreliable reviews that may cause inconsistency and ambiguity to readers. Hence, theoretical and managerial areas for social media analytics are enriched on the basis of online review meta-data in hospitality and tourism.     

An introduction to helpful forecasting methods for hotel revenue management

Revenue management is a key tool for hotel managers’ decision-making process. Cutting-edge revenue management systems have been developed to support managers’ decisions and all have as an essential component an accurate forecasting module. This paper aims to introduce new time series forecasting models to be considered as a tool for forecasting daily hotel occupancies. These models were developed in a state space modelling framework which is capable of tackling seasonal complexities such as multiple seasonal periods and non-integer seasonality. An empirical study was carried out to illustrate how a practitioner may apply and compare the performance of different models when forecasting a hotel’s daily occupancy. Results showed that the trigonometric model based on the new modelling framework generally outperformed the majority of the other models. These findings are potentially useful to the entire revenue management community facing the challenge of accurately forecasting a hotel’s daily demand.     

Prediction accuracy for reservation-based forecasting methods applied in Revenue Management

With a few notable exceptions, airlines and hospitality forecasting research has been focused so far on point predictions of customers’ bookings. However, Revenue Management decisions are subject to a much greater risk when based exclusively on point predictions. To overcome this drawback, we propose a stochastic framework that allows the construction of prediction intervals for reservation-based (pickup) forecasting methods, which are widely used in the industry. Moreover, we introduce an extension of the multiplicative pickup technique based on Generalized Linear Models. We test the proposed framework with real reservation data from a medium-sized hotel on Lake Maggiore (Italy) and we obtain more efficient prediction intervals relative to classical time series methods. Our approach can be useful to hotel revenue managers that wish to make more informed decisions, planning alternative pricing and room allocation strategies for a range of possible demand scenarios.     

Zero price effect on hotel demand: Evidence from a discrete choice experiment

A Choice Experiment is employed to analyze the effect of a free night promotion on hotel demand in the setting of a relatively underdeveloped area in China. Results from Error Components models show evidence in favor of a non-rational “zero price effect” (ZPE): with total price and all other aspects equal, people tend to choose the hotel which offers one free night. In addition, free pricing is shown to have stronger effects in diverting preferences than a trivial price (1 RMB). However, it is not the only successful psychological pricing strategy; its effects do not significantly differ from those of a materially equivalent discount. Building upon recent methodological innovations using Choice Experiments to study pricing strategy, this paper is the first to extend the technique to study the ceteris paribus ZPE. Our findings can help hotels make use of the ZPE to attract consumers.     

Forecasting Chinese cruise tourism demand with big data: An optimized machine learning approach

After more than ten years of exponential development, the growth rate of cruise tourist in China is slowing down. There is increasingly financial risk of investing in homeports, cruise ships and promotional activities. Therefore, forecasting Chinese cruise tourism demand is a prerequisite for investment decision-making and planning. In order to enhance forecasting performance, a least squares support vector regression model with gravitational search algorithm (LSSVR-GSA) is proposed for forecasting cruise tourism demand with big data, which are search query data (SQD) from Baidu and economic indexes. In the proposed model, hyper-parameters of the LSSVR model are optimized with GSA. By comparing these models with various settings, we find that LSSVR-GSA with selected mobile keywords and economic indexes can achieve the highest forecasting performance. The results indicate the proposed framework of the methodology is effective and big data can be helpful predictors for forecasting Chinese cruise tourism demand.     

Using deep learning and visual analytics to explore hotel reviews and responses

This study aims to use computational linguistics, visual analytics, and deep learning techniques to analyze hotel reviews and responses collected on TripAdvisor and to identify response strategies. To this end, we collected and analyzed 113,685 hotel reviews and responses and their semantic and syntactic relations. We are among the first to use visual analytics and deep learning-based natural language processing to empirically identify managerial responses. The empirical results indicate that our proposed multi-feature fusion, convolutional neural network model can make different types of data complement each other, thereby outperforming the comparisons. The visualization results can also be used to improve the performance of the proposed model and provide insights into response strategies, which further shows the theoretical and technical contributions of this study.     

A piecewise linear approach to modeling and forecasting demand for Macau tourism

In this paper, we construct and use a piecewise linear method to model and forecast, on a monthly basis, the demand for Macau tourism. Data over the period January 1991–December 2005 and a seasonally adjusted series for tourism demand are used. The study examines 4 forecasting horizons ranging from 6 to 24 months in advance. Mean absolute percentage errors and root mean square errors are adopted as criteria for evaluating the accuracy of the forecasting exercises. Finally, the forecasts of piecewise linear model are compared with those of autoregressive trend model, seasonal autoregressive integrated moving average and its arch-rival fractionally integrated autoregressive moving average models. The piecewise linear model is more accurate than the three benchmark models tested and the improvement is practically significant.     

The impact of the Asian financial crisis on Japanese demand for travel to Hong Kong: A study of various forecasting techniques

The Asian financial crisis has drawn worldwide attention because of its significant economic impact on local economics, especially on the economy of a tourism‐dependent destination. Unfortunately, there have been very few articles about the relationship of the Asian financial crisis and tourism demand forecasting. This relative lack of prior studies on the Asian financial crisis and tourism demand forecasting is particularly true in the context of Hong Kong. This article reports on a study that utilized officially published data to test the accuracy of forecasts of Japanese demand for travel to Hong Kong, measured in terms of the number of Japanese tourist arrivals. Seven commonly‐used tourism forecasting techniques were used to determine the forecasting accuracy. The quality of forecasting accuracy was measured in five dimensions. Experimental results indicated mixed results in terms of forecasting accuracy. Overall, artificial neural network outperformed other techniques in three of the five dimensions.     

Forecasting tourist arrivals with machine learning and internet search index

Previous studies have shown that online data, such as search engine queries, is a new source of data that can be used to forecast tourism demand. In this study, we propose a forecasting framework that uses machine learning and internet search indexes to forecast tourist arrivals for popular destinations in China and compared its forecasting performance to the search results generated by Google and Baidu, respectively. This study verifies the Granger causality and co-integration relationship between internet search index and tourist arrivals of Beijing. Our experimental results suggest that compared with benchmark models, the proposed kernel extreme learning machine (KELM) models, which integrate tourist volume series with Baidu Index and Google Index, can improve the forecasting performance significantly in terms of both forecasting accuracy and robustness analysis.     

Comparative study of deep learning models for analyzing online restaurant reviews in the era of the COVID-19 pandemic

Online reviews remain important during the COVID-19 pandemic as they help customers make safe dining decisions. To help restaurants better understand customers’ needs and sustain their business under current circumstance, this study extracts restaurant features that are cared for by customers in current circumstance. This study also introduces deep learning methods to examine customers’ opinions about restaurant features and to detect reviews with mismatched ratings. By analyzing 112,412 restaurant reviews posted during January-June 2020 on Yelp.com, four frequently mentioned restaurant features (e.g., service, food, place, and experience) along with their associated sentiment scores were identified. Findings also show that deep learning algorithms (i.e., Bidirectional LSTM and Simple Embedding + Average Pooling) outperform traditional machine learning algorithms in sentiment classification and review rating prediction. This study strengthens the extant literature by empirically analyzing restaurant reviews posted during the COVID-19 pandemic and discovering suitable deep learning algorithms for different text mining tasks.     

Customer purchase forecasting for online tourism: A data-driven method with multiplex behavior data

Online tourism has received increasing attention from scholars and practitioners due to its growing contribution to the economy. While related issues have been studied, research on forecasting customer purchases and the influence of forecasting variables, online tourism is still in its infancy. Therefore, this paper aims to develop a data-driven method to achieve two objectives: (1) provide an accurate purchase forecasting model for online tourism and (2) analyze the influence of behavior variables as predictors of online tourism purchases. Based on the real-world multiplex behavior data, the proposed method can predict online tourism purchases accurately by machine learning algorithms. As for the practical implications, the influence of behavior variables is ranked according to the predictive marginal value, and how these important variables affect the final purchase is discussed with the help of partial dependence plots. This research contributes to the purchase forecasting literature and has significant practical implications.