A MACHINE LEARNING APPROACH TO IMPROVING FORECASTING ACCURACY OF HOTEL DEMAND: A COMPARATIVE ANALYSIS OF NEURAL NETWORKS AND TRADITIONAL MODELS

Abstract

Demand forecasting is a key component of successful hotel operations and revenue management. In this research, we suggest a combined neural network method that incorporates heterogeneous data sets - time series and advance booking information along with seasonality components. We provide a comparative analysis to improve forecasting accuracy by investigating various forecasting methods including advance booking models, time series models. Our exhaustive study shows that the neural network approach outperforms traditional forecasting models overall. Moreover, we observe that the true behavior of daily demand may be a more complex phenomenon than single type data can capture, and that our combined neural network model may improve forecasting accuracy.