Multistep Prediction of Bus Arrival Time with the Recurrent Neural Network

Accurate predictions of bus arrival times help passengers arrange their trips easily and flexibly improve travel efficiency. Thus, it is important to manage schedule the buses for efficient deployment ease traffic congestion, which improves service quality public transport system. However, due many variables disturbing scheduled transportation, accurate prediction challenging. For time a bus, this research adopted recurrent neural network (RNN). prediction, affecting were investigated from data set containing route, driver, weather, schedule. Then, stacked multilayer RNN model was created with that categorized into four groups. The separate multi-input spatiotemporal sequence applied leaving all Shandong Linyi route. result simulation revealed convolutional long short-term memory (ConvLSTM) showed highest accuracy among tested models. propagation error number steps influenced accuracy.