Title: Individual Abnormal Travel Characteristics Recognition and OD Matrix Analysis based on AFC Data (Excerpt)
University: Chinese People's Public Security University
Instructor: Chen Peng
Team leader: Yin Wenrui
Team members: Xiao Song, Yao Minggang, Gu Haishuo
The development of the city depends on a perfect urban road network system. In the past, the planning of urban rail transit network was often made by experience, manual analysis and questionnaire survey. With bus IC cards popularization, it is more scientific and accurate to analyze residents' travel characteristics and passenger flow with AFC data in this era of cloud computing and big data.
However, with the continuous establishment of the urban public transport system, pick-pocketing on those transports is also increasing. More than 10 crimes may occur in one day on a bus route. Unfortunately, there are currently no effective prevention and control methods for such crimes. Public transport pick-pocketing often commits in the form of gangs, and according to police statistics, the occurrences have certain regularity. In this case, the use of AFC data could be effective in the prevention and control of those crimes, since it is able to reflect the passenger flow and individual travel characteristics using information technology. Finally, we can obtain patrol suggestions for pickpocket gangs.
In order to realize the effective identification of abnormal travel characteristics, this research firstly acquired and cleaned the data through Python. Next, we utilized Baidu API to distinguish working days, non-working days, peak periods, off-peak periods establishing a recommendation time matrix. Then we established rules to sift abnormal cards, and analyzed the reachable sites by elliptical model obtaining the patrol & prevention suggestions for key areas. Further, the rules to sift companion cards were established, and the OD matrix analysis and visualization were carried out to obtain the key areas of the criminal gang.
Count all enters and exits of companion cards, and record the number of occurrences of those stations to find its internal rules. Taking a criminal gang as an example, the gang has a relatively fixed entry and exit station, then it can be judged as locals-committing crimes. According to statistical analysis, the gang lives near Tongzhou Beiguan. Visualize the companion cards’ OD matrixes of two people in the gang separately, as the figure below shows. Then connect the two stations with an arc. The thickness of the arc is related to the number of times the card is swiped. The result shows that this group often stay at traffic-congested and crowded areas near Chegongzhuang West Station, Beijing West Railway Station and the Military Museum Station, which are in the vicinity of Tongzhou Beiguan. It verifies that the abnormal cards have a low degree of travel consistency. In addition, it can analyze the correlation between the stations and verify the proportional relation between the number of abnormal card occurrences and the passenger flow.
OD Track Visualization of Companion Cards
The work realizes the sifting of the abnormal cards and the companion cards through rules, and uses the OD matrix to analyze the accompanying abnormal travel characteristics. The SuperMap iClient 9D Beta for Leaflet is used to develop the OD matrix visualization toolbox. The work performs gray box trajectory analysis of those stations that we only know to enter and exit using the elliptical model. Further, it draws a Kernel density map and compares with the actual police statistics to verify the criminal gangs’ footholds. According to the hot lines and hot sites of the abnormal cards, the patrol & prevention suggestions for key areas are obtained. And by locating companion cards landing points, we can get arresting recommendation of the criminal gangs, which provides the policy a practical method for improving the pick-pocketing prevention and detection.
The work is for finding abnormal travel individuals from the perspective of bus card big data analysis, and then identifies and predicts potential criminals. It is a very typical public security big data intelligence analysis. Based on the comprehensive application of various analysis tools, the team boldly innovates from previous methods and realizes the effective mining of public transport big data. The research process is scientific, standardized, and reliable, not only has a high academic significance but also has high application value in the actual work of public security.
© 1997-2020 SuperMap Software Co., Ltd.
6/F, Building 107, No. A10, Jiuxianqiao North Road, Chaoyang District, Beijing