Researchers test using smart card travel ticket data to find pickpockets

In a presentation given at the Knowledge Discovery and Data Mining conference in 2016, researchers discussed a method of using the data generated by smart card public transport tickets to catch pickpockets. In a study of 6 million passenger movements in Beijing, the researchers used a classifier to pick out anomalous journeys - sudden variations in the patterns of ordinary travellers or routes that made no sense. A second classifier primed with information derived from police reports and social media posts identifying hotspots for pickpocketing then tried to spot the pickpockets among the anomalous journeys. The system identified 93% of the pickpockets caught by police during the period in question; however, it identified 14 travellers it deemed suspicious for every known pickpocket. Transport for London's chief technology officer expressed scepticism on the basis that his experience shows that people make all sorts of journeys for all sorts of reasons. The researchers, however, intend to extend their work to see if they can identify alcoholics, drug users, homeless people, and drug dealers.

Writer: The Economist
Publication: The Economist
Publication date: 2016-08-18