The accurate short-term passenger flow prediction is of great significance for real-time public transit management, timely emergency response as well as systematical medium and long-term planning. In this paper, we propose an end-to-end deep learning framework that can simultaneously make multi-step predictions for all stations in a large scale metro system. A sequence to sequence model embedded with the attention mechanism forms the backbone of this framework. The sequence to sequence model consists of an encoder network and a decoder network, making it good at modeling sequential data with varying lengths and the attention mechanism further enhances its ability to capture long-range dependencies. We use the proposed framework to predict the number of passengers alighting at each station in the near future, given the number of passengers boarding at each station in the last few short-term periods. The large quantities of real-world data collected from Singapore’s metro system are used to validate the proposed model. In addition, a set of comparisons made among our model and other classical approaches evidently indicates that the proposed model is more scalable and robust than other baselines in making multi-step and network-wide predictions for short-term passenger flow.

Sequence to sequence learning with attention mechanism for short-term passenger flow prediction in large-scale metro system

Abstract The accurate short-term passenger flow prediction is of great significance for real-time public transit management, timely emergency response as well as systematical medium and long-term planning. In this paper, we propose an end-to-end deep learning framework that can simultaneously make multi-step predictions for all stations in a large scale metro system. A sequence to sequence model embedded with the attention mechanism forms the backbone of this framework. The sequence to sequence model consists of an encoder network and a decoder network, making it good at modeling sequential data with varying lengths and the attention mechanism further enhances its ability to capture long-range dependencies. We use the proposed framework to predict the number of passengers alighting at each station in the near future, given the number of passengers boarding at each station in the last few short-term periods. The large quantities of real-world data collected from Singapore’s metro system are used to validate the proposed model. In addition, a set of comparisons made among our model and other classical approaches evidently indicates that the proposed model is more scalable and robust than other baselines in making multi-step and network-wide predictions for short-term passenger flow.

Sequence to Sequence Learning with Attention Mechanism for Short-term Passenger Flow Prediction in Large-scale Metro System

Traffic state prediction is a key component in intelligent transport systems (ITS) and has attracted much attention over the last few decades. Advances in computational power and availability of a large amount of data have paved the way to employ advanced neural network (NN) models for ITS, including deep architectures. There have been various NN‐based approaches proposed for short‐term traffic state prediction that are surveyed in this article, where the existing NN models are classified and their application to this area is reviewed. An in‐depth discussion is provided to demonstrate how different types of NNs have been used for different aspects of short‐term traffic state prediction. Finally, possible further research directions are suggested for additional applications of NN models, especially using deep architectures, to address the dynamic nature in complex transportation networks.

Survey of neural network-based models for short-term traffic state prediction

The paper points to a multi-criteria decision-making model based on the rough set theory application. The model demonstrates exceptional importance of the software application of the rough sets to decision-making in the security forces operations. Applying the rough sets represents a useful tool when the data, needed for the decision-making process, include vagueness and uncertainty. By applying the model based on the applicative use of the rough sets, specific decision-making rules are formulated. These rules guide the decision-makers through the complete process of planning the security operations

A multi-criteria decision-making (MCDM) model in the security forces operations based on rough sets

Mobility is one of the major dimensions of smart city design and development. Transportation analysis and prediction play important parts in mobility research and development. Recent years have seen many new types of transportation data emerging, such as social media and Global Positioning System (GPS) data. These data contains hidden knowledge, which can be used in many applications to improve city operations; road traffic prediction is one aspect of this. Researchers have traditionally used single traffic flow prediction methods, which work well only under specific conditions. Some work has emerged in recent years on combining these methods into various hybrid methods. However, this work is in its infancy, and further investigations are required. More importantly, these hybrid methods have mostly been developed on stand-alone, nondistributed platforms, limiting the data and problem sizes that can be addressed, as well as the accuracy that can be achieved. This chapter gives a review of traffic flow prediction and modeling methods and discusses the limitations of each method. A review of the various types of transportation traffic data sources is provided. Notable big data analysis tools, including the Apache Spark platform, are described. Finally, we describe a hybrid method for road traffic prediction and provide a tutorial on the process of hybrid traffic flow prediction. The hybrid method is based on the autoregressive integrated moving average (ARIMA) and support vector machine (SVM) methods.

Hybrid Statistical and Machine Learning Methods for Road Traffic Prediction: A Review and Tutorial

With the successive increase in usage of vehicles, severe traffic congestion is on the rise. This in turn leads to increase in environmental pollution and accidents which ultimately affects the safety, time consumed and money spent of the transport users. The solution to this critical problem is traffic flow prediction depending on which traffic control measures and traffic management can be done. Short term traffic flow prediction has become one of the hot topics of research in intelligent transportation system (ITS). It predicts the traffic flow count in the next time interval that usually varies in the range of five minutes to half an hour. This work presents the application of Support vector machine (SVM) for predicting the traffic flow in a short span. Highway data of midblock segment collected from Chennai, India is used for the analysis. Improved performance measure shows satisfactory results upon application of SVM.

Performance analysis of support vector machine for traffic flow prediction

Short term traffic flow prediction has become one of the important research fields in intelligent transportation system. The prediction of this traffic flow information quickly and accurately is important for traffic control and guidance to initiate the measuring steps well in advance. It makes the transport users better informed and makes the transport network smarter, safer and more coordinated. It plays a crucial role in individual dynamic route guidance, advance traffic information system (ATIS) and advance traffic management system (ATMS). This paper discusses the implementation of traffic flow prediction model using support vector machine. Rough set is used as a post processing tool to validate the prediction result. The objective is to improve traffic flow prediction performance. Data near Perungudi toll plaza in IT corridor in Chennai, India is used for the analysis. It is found that the use of rough set results in satisfactory performance improvement which is evaluated using mean square error as the performance measures. General Terms Intelligent Transportation Systems, Soft Computing Methods.

/pdf/performance-improvement-of-traffic-flow-prediction-model-1dfduqmuwf.pdf

Performance Improvement of Traffic Flow Prediction Model using Combination of Support Vector Machine and Rough Set

As traffic demand is growing rapidly, traffic congestion has become a severe problem everywhere in the world. Accurate prediction of traffic flow is therefore essential for the transport users to make smarter and effective decisions like route guidance, mode of transport, time of travel etc. Short term traffic flow prediction has become one of the important research fields in intelligent transportation system (ITS) and forms crucial base for the traffic management, traffic guidance and control strategy. This work suggests the application of a neuro-fuzzy hybrid system which brings together the complementary capabilities of both neural networks and fuzzy logic for short term traffic flow prediction application. The objective is to improve prediction accuracy. Data from highway of Chennai, India is used for the analysis. The use of hybrid system results in satisfactory improvement in the performance measure.

Minal Deshpande

Papers

Performance analysis of support vector machine for traffic flow prediction

Performance Improvement of Traffic Flow Prediction Model using Combination of Support Vector Machine and Rough Set

Short term traffic flow prediction based on neuro-fuzzy hybrid sytem