A new unconstraining method for demand forecasting

In this paper we propose a new unconstraining method for demand forecasting. Since true demand forecasting is a key aspect of hotel room revenue management systems, inaccurate forecasts will significantly impact the performance of these systems. We propose a method based on a Monte Carlo simulation forecasting model and an Expectation-Maximization (EM) algorithm that, unlike traditional statistical unconstraining methods, takes into account the time-series aspect of the observed demand (trend and seasonality) and handles the complex distribution of the demand and its relationship with other system variables/parameters. Our approach is presented in an efficient and simple algorithm. We considered as a case study the data for Plaza Hotel, Alexandria, Egypt. The primary results of our approach demonstrate that it can produce efficient solutions for estimating the unconstrained demand which provide revenue improvements in the industry.