Award Date


Degree Type


Degree Name

Doctor of Philosophy (PhD)


William F. Harrah College of Hospitality

First Committee Member

Amanda Belarmino

Second Committee Member

Ashok Singh

Third Committee Member

Carola Raab

Fourth Committee Member

Kaushik Ghosh

Number of Pages



The aim of this dissertation is to create and test a risk induced game-theoretic price forecasting model. The models were tested with datasets from 3 Upper Midscale hotels in 3 locations (urban, interstate and suburb), one hotel from each location. The data was obtained from STR, a leading hospitality marketing company which consolidates all of the daily hotel data from hotels in the United States. Multiple error measures were used to compare the accuracy of models. Three LSTM models were proposed and tested; LSTM model 1 that relied on ADR to forecast ADR, LSTM model 2 that used ADR, supply, demand, and day of the week to generate the forecast, and finally LSTM model 3 that used all the predictors of LSTM model 2 plus ADR of 4 competitors of the same size and scale to predict ADR values. The LSTM models were tested against traditional forecasting methods. The findings showed that LSTM model 2 was the most accurate of all the models tested. Moreover, LSTM model 1 and 3 showed higher accuracy than traditional models in some cases. In particular, all the LSTM models outperformed the traditional methods in the most volatile property (property C). Overall, the results indicated the higher accuracy of LSTM models for times of uncertainty. Finally, estimation of Value at Risk was introduced into the LSTM models, however the accuracy of the models did not change significantly.


Artificial Intelligence; Hospitality Industry; Machine Learning; Neural Networks; Price Forecasting; Revenue Management


Business Administration, Management, and Operations

File Format


File Size

7500 KB

Degree Grantor

University of Nevada, Las Vegas




IN COPYRIGHT. For more information about this rights statement, please visit