Short-term Railway Passenger Demand Forecasting

Forecasting passenger arrival is crucial for daily
operations. In revenue management, predicting the
number of passengers at departure offers essential
information for seat allocation, overbooking, and
pricing decisions. In recent years, Artificial
Neural Networks have been successfully applied on
solving time series forecasting problems. In this
study, we show how to design ANN models to predict
short-term railway passenger demand by using input
information as effective as possible. The concept of
divide-and-conquer is utilized in designing new
structures in this study; three novel networks
termed multiple temporal units neural network,
parallel ensemble neural network and input recurrent
neural network are proposed. Furthermore, six
related issues are tested to show the predictive
capability of individual models and their
combinations. The book should shed some light on ANN
network structures and also the benefit of combining
models within ANN and between various methodologies;
it should be useful for researchers and
practitioners who are in the field of time series
forecasting, ANN, revenue management and railway
transportation.

Autorentext

Tsung-Hsien Tsai is currently a research associate at the School of Hotel Administration, Cornell University. He got his Ph.D. in Transportation and Communication Management Science from National Cheng Kung University in Taiwan. His research interests are time series forecasting, revenue management and Artificial Neural Networks.

Klappentext

Forecasting passenger arrival is crucial for daily operations. In revenue management, predicting the number of passengers at departure offers essential information for seat allocation, overbooking, and pricing decisions. In recent years, Artificial Neural Networks have been successfully applied on solving time series forecasting problems. In this study, we show how to design ANN models to predict short-term railway passenger demand by using input information as effective as possible. The concept of divide-and-conquer is utilized in designing new structures in this study; three novel networks termed multiple temporal units neural network, parallel ensemble neural network and input recurrent neural network are proposed. Furthermore, six related issues are tested to show the predictive capability of individual models and their combinations. The book should shed some light on ANN network structures and also the benefit of combining models within ANN and between various methodologies; it should be useful for researchers and practitioners who are in the field of time series forecasting, ANN, revenue management and railway transportation.