Research Study: Data Preprocessing Using Machine Learning for Prediction of Booking Cancellations

Cancellations of reservations have a substantial effect on the choices that hospitality businesses make about management control. In order for hotels to minimize the blow of guest cancellations, this report have instituted stringent cancellation rules and scheduling strategies. However, these strategies may have a detrimental effect not only on income but also on the hotel’s image. A machine learning-based system model was designed in order to lessen the severity of this effect. It does this by using the data from the hotel’s Management Company Systems and training a classification algorithm on a daily basis. This allows it to determine which reservations are “likely to cancel” and, as a result, compute net demands. This prototypes, which has been installed in a production setting at two hotels, allows the assessment of the effect of measures made to respond upon bookings that have been projected as likely to cancel by implementing A/B testing. Results reveal strong prototype effectiveness and provide crucial indicators for the development of research while also providing evidence that reservations approached by hotels cancel less often than bookings that are not approached by hotels.