Published Paper Details:

Huge data acquired by automated fare collection (AFC) systems allows for the analysis of both personal and group mobility patterns in urban regions. Previous research on AFC data has mostly concentrated on determining passenger movement patterns. However, we used such data in a novel way to identify pickpocket suspects. Stopping pickpockets on public transportation is critical for increasing passenger happiness and public safety. Nonetheless, distinguishing thieves from ordinary travellers is difficult in practise. We created a suspect identification and surveillance system in this research that can identify pickpocket suspects based on their daily transit data. Specifically, we collected a number of important information from each passenger's daily transit system activity. we used the strengths of unsupervised outlier identification and supervised classification models to identify robbers, who often exhibit anomalous travelling behaviours. The efficiency of our strategy was demonstrated by experimental data. We also created a prototype system for use by security personnel.