Estimating truncated hotel demand: A comparison of low computational cost forecasting methods
Abstract
The aim of this thesis is to evaluate the effectiveness of six selected low computational cost hotel demand forecasting methods (SA, SMA, EMA, DEMA, BP and PU) in terms of restoring truncated demand data, and then identify a low-cost and easy to follow demand forecasting method that can be used by U.S. independent hotels. Obtaining revenue gains by applying demand forecasting techniques have been proved by many studies in hospitality and other related industries. However, few studies have focused on low computational forecasting methods' comparison in hospitality field. For this reason, the author decided to test the performance of six selected demand forecasting techniques, with the aim of identifying an effective method for hotels operators constrained by financial resources and expertise. This thesis first simulates leisure and business real demand booking curves under a pre-decided increasing rate in each of three leisure/business ratio scenarios (1:3, 1:1, and 3:1). In the second stage, true demands are truncated in three cases. They are 1) capacity truncation, 2) 50% truncation of total business demand, and 3) 25% truncation of total business demand. And then, six selected forecasting methods are applied to the truncated demand. Finally, the forecasting accuracy for each method is evaluated in both statistical and economical models. The results of the experiment indicate that PU method outperform all the other selected methods and was proved to be the most effective forecasting method for U.S. independent hotels. Other new findings include that the data restoration accuracy ranged from a negative relationship with the business demand proportion of total bookings, and the higher the percentage the business bookings were truncated, the smaller the detruncation error occurs. The results also shows that the less the business booking was truncated; the more variable the forecasting error occurs. An interesting finding of this thesis is that in some specific circumstances, the results of statistical evaluation do not completely in accordance with economical evaluation results.
Degree
M.S.
Advisors
Tang, Purdue University.
Subject Area
Management|Commerce-Business|Recreation
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.