Predicting hotel booking cancellations to decrease uncertainty and increase revenue

Abstract

Booking cancellations have a substantial impact in demandmanagement decisions in the hospitality industry. Cancellations limit the production of accurate forecasts, a critical tool in terms of revenue management performance. To circumvent the problems caused by booking cancellations, hotels implement rigid cancellation policies and overbooking strategies, which can also have a negative influence on revenue and reputation.Using data sets from four resort hotels and addressing booking cancellation prediction as a classification problem in the scope of data science, authors demonstrate that it is possible to build models for predicting booking cancellations with accuracy results in excess of 90%. This demonstrates that despite what was assumed by Morales and Wang (2010) it is possible to predict with high accuracy whether a booking will be canceled.Results allow hotel managers to accurately predict net demand and build better forecasts, improve cancellation policies, define better overbooking tactics and thus use more assertive pricing and inventory allocation strategies.Keywords: Data science, hospitality industry, machine learning, predictive modeling, revenue management. ResumoO cancelamento de reservas tem um impacto substancial nas decisões de gestão da procura na industria hoteleira. Os cancelamentos limitam a produção de previsões precisas, uma ferramenta crítica em termos de desempenho de gestão da receita. Para limitar os problemas causados pelo cancelamento de reservas, os hotéis implementam políticas de cancelamento rígidas e estratégias de overbooking, as quais podem vir a ter influência negativa sobre a receita e reputação social. Usando conjuntos de dados de quatro hotéis de resort e abordando a previsão de cancelamento de reservas como um problema de classificação no âmbito da Data Science, os autores demonstram que é possível construir modelos para prever cancelamentos de reservas com resultados superiores a 90%. Estes resultados permitem demonstrar que apesar do que foi assumido por Morales e Wang (2010) é possível prever com alta precisão se uma reserva será cancelada. Os resultados permitem que os hoteleiros prevejam com melhor precisão a procura líquida e construam melhores previsões, melhorem as políticas de cancelamento, definam melhores táticas de overbooking e usem estratégias de alocação de inventário com preços mais assertivos.

Big Data in Hotel Revenue Management: Exploring Cancellation Drivers to Gain Insights Into Booking Cancellation Behavior

“…Therefore, bookings cancellation forecast/prediction that uses data representing a large number of these factors is likely to present better performance results. This may help explain the results obtained by Antonio et al (2017aAntonio et al ( , 2017bAntonio et al ( , 2017c, Falk & Vieru, 2018) and Huang et al (2013) Antonio et al (2017aAntonio et al ( , 2017bAntonio et al ( , 2017c added another feature, the customer's previous cancellation history, which represents another known cancellation factor (C.-C. Chen, 2016;C.-C. Chen et al, 2011;Talluri & Van Ryzin, 2005). Nevertheless, all of these features were obtained from the same source, the PMS.…”

“…Second, previous studies that employed machine learning algorithms draw conclusions from prediction error results obtained from validation sets built with data from the same period of the training data (Antonio et al, 2017a(Antonio et al, , 2017c(Antonio et al, , 2017cHuang et al, 2013). By creating a test set consisting of bookings from a period that was not included in the training and validation sets, we demonstrated that models that produce good results with known data do not always generalize well.…”

Modelling the cancellation behaviour of hotel guests

“…Based on 240,000 booking records, Morales and Wang (2010) find that several guest-, booking-and room-specific characteristics are relevant in forecasting cancellation rates. Antonio, de Almeida, and Nunes (2017) show that booking channel, arrival month, room type, booking lead time, and country of origin are the most important predictors.…”

Predictive models for hotel booking cancellation: a semi-automated analysis of the literature

“…Azadeh et al (2013) developed a model to predict bookings and cancellations for a railway operator. Antonio et al (2017aAntonio et al ( , 2017b employed hotel data to develop models to predict each booking's cancellation probability, and, simultaneously, net demand. Tsai (2011) used railway data to develop a model to forecast the cancellation rate.…”

“…However, such policies can have adverse effects in terms of reallocation costs, social reputation, or revenue decrease due to the discounts granted (Guo, Dong, & Ling, 2016;Noone & Lee, 2010). To reduce the negative impact of overbooking and restrictive cancellation policies, as well as the uncertainty in demand management decisions, cancellations and no-show forecasts are used as inputs in RMSs (Antonio, Almeida, & Nunes, 2017a;Morales & Wang, 2010;Talluri & Van Ryzin, 2005). However, despite the importance of booking cancellation forecast models (Chen, 2016), only a few works have tested or developed forecast cancellation models.…”