Leveraging Intra-Period and Inter-Period Features for Enhanced Passenger Flow Prediction of Subway Stations

TL;DR

A novel model called MPSTN is proposed that leverages features from different periods by transforming one-dimensional time series data into two-dimensional matrices based on periods that incorporates a CNN module to extract temporal information from different periods and a graph neural network module to integrate spatial information from different stations.

Abstract

Accurate short-term passenger flow prediction of subway stations plays a vital role in enabling subway station personnel to proactively address changes in passenger volume. Despite existing literature in this field, there is a lack of research on effectively integrating features from different periods, particularly intra-period and inter-period features, for subway station passenger flow prediction. In this paper, we propose a novel model called \textbf{M}uti \textbf{P}eriod \textbf{S}patial \textbf{T}emporal \textbf{N}etwork \textbf{MPSTN}) that leverages features from different periods by transforming one-dimensional time series data into two-dimensional matrices based on periods. The folded matrices exhibit structural characteristics similar to images, enabling the utilization of image processing techniques, specifically convolutional neural networks (CNNs), to integrate features from different periods. Therefore, our MPSTN model incorporates a CNN module to extract temporal information from different periods and a graph neural network (GNN) module to integrate spatial information from different stations. We compared our approach with various state-of-the-art methods for spatiotemporal data prediction using a publicly available dataset and achieved minimal prediction errors. The code for our model is publicly available in the following repository: https://github.com/xiannanhuang/MPSTN

Figures (5)

• Figure 1: The inflow of a station in the past seven days. The data in the red area represents the inflow of several time intervals immediately preceding the future time to be predicted, while the data in the grey area represents the inflow at the same time interval to be predicted on several previous days.
• Figure 2: Analysis process of multi-period data. a) raw inflow data of a station in 7 days. b) folding the inflow data according to the period. c) folding data as a matrix showing structural features similar to photos.
• Figure 3: The workflow of the first category of methods. Each node in the bottommost column represents a traffic unit, such as a subway station or a city area. The nodes and edges can be considered as a graph, which reflects the spatial information of the traffic network.
• Figure 4: Different Spatial-temporal Graphs. a) represents graph with connections between the same node at adjacent time intervals. b) represents graph with connections not only between the same node at adjacent time intervals but also between adjacent nodes at adjacent time intervals. To ensure clarity in the visualization, only the cross-time connections of the central node and in time 2 are depicted.
• Figure 5: Workflow of our appeared methods. The left part represents the graph and the right part of the figure represent the note feature.