Origin and destination forecasting on dockless shared bicycle in a hybrid deep-learning algorithms

Abstract

Nowadays, the dockless shared bicycle has a positive influence on people’s travel, thus it is useful to analyze the spatio-temporal features of shared bike. Due to the limitations of CNN or LSTM, the spatial correlation and time dependence is inferior to capture. In this paper, a combination of CNN and LSTM named CLTFP in deep learning model is applied to predict the travel distance and OD distribution of shared bicycles under different conditions of time and space. Experiments show that CLTFP has better performance to capture spatiotemporal correlations.