The Improvement of Revenue Management in the Hoteling Industry using Neural Networks to Determine Stochastic Parameter in an Overbooking Model

The use of revenue management models has been increased in various industries. The cause of such increasing is as a result of performance and profitability of these models in businesses. Hoteling industry is considered as an important business in the field of revenue management that has a reservation process and stochastic variables due to it. Classic overbooking model is considered as a common model in revenue management that causes to make a trade-off between the number of present customers and no-show customers. This model makes a situation for studying the functions which describe costumers’ presence distribution in probable form and then we can add some customers to system for increasing revenue due to no-shows. In this research, the binomial probability distribution using in overbooking model has been improved and estimated its probable parameter more accurately using artificial neural network as a tool in no-show estimation. This estimation is caused by fitting to effective indexes in show-up or no-show process using one-layer or multi-layer perceptron neural network. Therefore, a dynamic model for each sale and customers’ reservation is represented that it can estimate the probability parameter of customers’ show-up or no-show considering effective indexes.