A Data-Driven Feature Based Learning Application to Detect Freeway Segment Traffic Status Using Mobile Phone Data

Abstract

With the finishing of the construction of the main body of a freeway network, adequately monitoring the traffic status of the network has become an urgent need for both travelers and transportation operators. Various methods are proposed to collect traffic information for this purpose. In this article, a data-driven feature-based learning application is implemented to detect segment traffic status using mobile phone data, building on the practical success of deep learning models in other fields. The traffic status estimation is achieved via the application of a three-level long, short-term memory model. Two phone features are extracted from the raw mobile phone data. A large-scale field experiment was conducted using actual data in Jiangsu, China collected over the “National Holiday Golden Week” of 2014. To evaluate the performance, both precision and recall scores are given along with the overall accuracy. The final results of the large-scale experiment indicate that the proposed application performed well and can be an emerging solution for traffic state monitoring when only limited roadside sensing devices are installed.

Classification of the Traffic Status Subcategory with ETC Gantry Data: An Improved Support Tensor Machine Approach

“…(5) While iteration termination conditions not meet do (6) Unlabeled data input to ISTM model trafc state classifer; (7) Select the classifcation tensor samples with high confdence and categorize them as input label samples; (8) Multiscale training of ISTM model according to input label samples; (9) Obtaining an ISTM model trafc state classifer; (10) Obtain tensor sets with trafc status labels T � (X 1 , y 1 ), (X 2 , y 2 ), ..., (X l , y l ) 􏼈 􏼉 End ALGORITHM 2: Te self-training process of the ISTM model. trafc fow for fve days.…”

“…Trafc state identifcation is afected by the quality of research data and model performance to some specifc-the trafc state subcategory research challenges the trafc data and models used. (1) Te collection of highway trafc data may rely on commonly used detectors, such as magnetic detectors, video detectors, and mobile phone data [8]. Due to the limitation of detection ability (e.g., vehicle type detection and vehicle diversion detection) and low deployment density of the above detectors, the quality of trafc data obtained is difcult to guarantee.…”

Mobile Phone Data Feature Denoising for Expressway Traffic State Estimation

“…Their experimental results suggested that the proposed method achieved a classification accuracy of 91% for low, medium, and high (three-level traffic states) traffic levels. Using feature extraction from raw mobile phone data and a three-level-long short-term memory model, Qiang Liu et al [9] conducted large-scale field experiments on actual data collected during the 2014 "Golden Week" in Jiangsu Province, China. The proposed application performed well and became an emerging solution for traffic status monitoring with limited roadside sensing equipment.…”