A user behavior-based ticket sales prediction using data mining tools: An empirical study in an OTA company

Traditional OTA (Online Travel Agent) is challenged by the Internet and mobile business with the evolution of information. The precise forecasting of ticket sales in OTA companies is beneficial to budget control and service quality. The paper develops an integrated forecasting model by combining the internal factors immediately influencing the ticket sales and the external factors reflecting the ticket sales market. The internal factors are selected such as the number of calling in certain duration, while the external factors include the attention of relevant search engine query data. After several key features are extracted using feature selection model, the machine learning algorithms can get the more accurate prediction, in contract to the basic experiments to explore the inherent rule of the sales data itself. Our proposed user behavior-based prediction model provides a feasible and efficiency tool for ticket sales prediction.