Deep Bidirectional and Unidirectional LSTM Neural Networks in Traffic Flow Forecasting from Environmental Factors

Abstract

The application of deep learning techniques in several forecasting problems has been increased the last years, in many scientific fields. In this research, a deep learning structure is proposed, composed mainly of double Bidirectional Long Short-Term Memory (Bi-LSTM) Network layers, for the prediction of the traffic flow in the study area. Also, traffic flow-related environmental factors were taken into consideration in order to construct the deep learning forecasting model. The final results have showed an increased accuracy of the proposed deep learning Bi-LSTM – based model compared to other machine learning models that were tested such as unidirectional LSTM networks, Support Vector Machines and Feedforward Neural Networks. © 2021, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.