Compound Prediction Research Based On Revenue Management

With the growing scale of the global tourism market, global distribution system(GDS) are increasingly infiltrated into all areas of people’s life, such as transport, accommodation, catering and entertainment. The competition of tourism market is fiercer than ever before. In accordance with the commitment of China’s opening to the outside world, the GDSs of foreign such as Amadeus, Galieo, Sabre and Wordspan are gradually enter the Chinese market. The research and application of revenue management theory promote the aviation and tourism industries in Europe and the US a qualitative leap in the management and operation level, and gradually become the core competitiveness of enterprises. There few applications of the revenue management in domestic GDS. In order to deal with the fierce competitive enviroment, to build our own revenue management system is imminent.Customer demand forecasting is the basis for the revenue management. This paper first reviews the history of the development of revenue management, and then summarizes the current popular revenue management prediction algorithms, such as moving average, exponential smoothing, and incremental method, and analyzed the characteristics of these algorithms. Most algorithms that based on the determined mathematical model can not describe the random behavior of the customer in the real world, and therefore has a large error. Based on this, this paper proposes a compound algorithm that is based on Monte Carlo simulation method and Holt exponential smoothing, and by means of randomized trials to simulate the random behavior of customers, thus effectively solves the randomness of the customer. In addition, the composite prediction algorithm can predict the number of days of customer stay and group bookings, which is not available in the other algorithms.Finally, this paper studies the hotel distribution platform, the test data is from the Dalian agent hotel booking record for2010-2012. This paper realizes the moving average method, the exponential smoothing, the incremental method and the proposed method on the same data set. Computer simulation results show that the compound method has the highest prediction accuracy and stability, the mean absolute percentage error (MAPE) of is19%, while the other prediction algorithm MAPE is more than21%.