Publicação

From Demand Forecasts to Pricing Decisions: An Interactive Hotel Revenue Management Simulator

Ver documento

Detalhes bibliográficos
Resumo:Hotels operate with perishable capacity and demand that varies strongly over time, which makes pricing decisions uncertain and dependent on reliable demand expectations. Integrated simulation environments that combine demand forecasting and dynamic pricing are still relatively rare, despite their potential value for decision support and training. This study develops an interactive hotel revenue management simulator that integrates data upload, baseline demand forecasting, and dynamic pricing within a single workflow, enabling risk-free scenario testing of price changes and revenue-oriented recommendations for selected stay dates. Calendar and event indicators are engineered as model inputs for baseline daily demand forecasting. Baseline demand is predicted using a gradient boosting regression model trained on a variance-stabilized target. Pricing outputs are derived by combining the baseline forecast with a calibrated demand-response formulation expressed in relative prices, using typical-price reference values and hotel-level price sensitivity estimates. Recommended prices are obtained through a bounded grid search that maximizes model-implied expected revenue. An empirical backtesting design with a time-based holdout period is used to assess forecasting accuracy and pricing performance against transparent baseline methods. The results from a Portuguese hotel booking dataset indicate that the forecasting component improves predictive accuracy relative to calendar-based baselines and that the pricing component yields higher model-implied expected revenue than simple median-based pricing rules under the simulator’s assumptions. The study contributes an integrated and reusable simulator artifact for hotel revenue management and offers practitioners a safe environment for pricing experimentation on their own booking data.
Autores principais:Schneider, Jan-Louis
Assunto:Hotel Revenue Management Revenue Management Simulator Optimal Pricing Dynamic Pricing Data-Driven Decision Making
Ano:2026
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso embargado
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:Hotels operate with perishable capacity and demand that varies strongly over time, which makes pricing decisions uncertain and dependent on reliable demand expectations. Integrated simulation environments that combine demand forecasting and dynamic pricing are still relatively rare, despite their potential value for decision support and training. This study develops an interactive hotel revenue management simulator that integrates data upload, baseline demand forecasting, and dynamic pricing within a single workflow, enabling risk-free scenario testing of price changes and revenue-oriented recommendations for selected stay dates. Calendar and event indicators are engineered as model inputs for baseline daily demand forecasting. Baseline demand is predicted using a gradient boosting regression model trained on a variance-stabilized target. Pricing outputs are derived by combining the baseline forecast with a calibrated demand-response formulation expressed in relative prices, using typical-price reference values and hotel-level price sensitivity estimates. Recommended prices are obtained through a bounded grid search that maximizes model-implied expected revenue. An empirical backtesting design with a time-based holdout period is used to assess forecasting accuracy and pricing performance against transparent baseline methods. The results from a Portuguese hotel booking dataset indicate that the forecasting component improves predictive accuracy relative to calendar-based baselines and that the pricing component yields higher model-implied expected revenue than simple median-based pricing rules under the simulator’s assumptions. The study contributes an integrated and reusable simulator artifact for hotel revenue management and offers practitioners a safe environment for pricing experimentation on their own booking data.