EMD-LSTM based deep learning inbound and outbound passenger flow prediction

As the subway has become one of the major modes of travel for people, the demand for rail transportation has increased. With effective algorithms and models, predicting the number of people entering and exiting the station in the future can play a crucial role for the railroads to regulate the trains in a reasonable way. In this paper, the real data of 28 days are decomposed by EMD algorithm, and multiple IMF components and one Res residual are decomposed. The decomposed IMF components are sequentially analyzed with the original data by Spearman correlation analysis and Kendall correlation analysis, and the IMF components with higher correlation are selected. The selected IMF components are trained together with the original data by LSTM model to derive the predicted passenger flow for the following week.