Technical gazette, Vol. 28 No. 2, 2021.
Preliminary communication
https://doi.org/10.17559/TV-20200513124207
Route Restoration Method for Sparse Taxi GPS trajectory based on Bayesian Network
Guangyao Li
; 1) Ningbo University, Ningbo City, Zhejiang Province, China 2) Jiangsu Province Collaborative Innovation Center for Modern Urban Traffic Technologies, Nanjing City, Jiangsu Province, China
Zhengfeng Huang*
; 1) Ningbo University, Ningbo City, Zhejiang Province, China 2) Jiangsu Province Collaborative Innovation Center for Modern Urban Traffic Technologies, Nanjing City, Jiangsu Province, China
Leyi Lou
; Zhejiang Sci-Tech University, Hangzhou City, Zhejiang Province, China
Pengjun Zheng
; 1) Ningbo University, Ningbo City, Zhejiang Province, China 2) Jiangsu Province Collaborative Innovation Center for Modern Urban Traffic Technologies, Nanjing City, Jiangsu Province, China
Abstract
In order to improve the availability of taxi GPS big data, we restore the chosen route for the sparse taxi GPS trajectory in this work. A trajectory restoration method based on Bayesian network is proposed. Compared with the traditional research solely based on time-spatial variables, this method additionally considers the characteristics of empty/heavy taxi status, weather conditions, drivers, vehicle running and other factors to carry out route restoration. A field case of grid network in Ningbo is taken to verify the applicability of the method, using the taxi GPS trajectory data from Ningbo Taxi Information Management Platform. The case results show that the accuracy of Bayesian network method based on multiple factors reaches 91.4%. Its performance is superior to the Multivariate logistic regression model. In addition, the proposed method is especially suitable for scenarios with a high missing rate of track data, such as a scene with timespan of about 5 min between neighbour trajectories.
Keywords
Bayesian network; missing rate; multiple factors; sparse Taxi GPS data; trajectory restoration
Hrčak ID:
255885
URI
Publication date:
17.4.2021.
Visits: 1.344 *