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.