Challenges in Measuring Fare Evasion
Fraud in public transportation is a major issue that causes significant financial and operational challenges for public transport operators. However, accurately measuring the extent of fare evasion is a challenging task, as current methods of measuring fare evasion have significant limitations. The good news is that artificial intelligence (AI) can help.
Current Fare Evasion Measurement Methods
The mass transit sector currently uses four main methods to measure fraud:
- Manual surveys: In these surveys, trained observers visually inspect passengers on board vehicles or at stations to detect fare evasion. The method is frequently time-consuming, biased or inaccurate, and can only capture a limited number of passengers at any given time.
- Ticket inspections: This involves transport staff checking passengers’ tickets or passes at random or predetermined intervals to estimate fare evasion. The method is susceptible to errors in sampling: for instance, there are significant differences in the fare evasion rates calculated by uniformed and plainclothes fare inspectors. Moreover, it is easily avoided when inspections are static, such as blanket controls in an underground corridor, particularly now that social media is widely used.
- Self-reporting surveys: This involves transportation operators surveying passengers about their fare payment behaviour, and using the results to estimate the level of fare evasion. Because passengers may be reluctant to admit fare evasion, the method is prone to bias or response inaccuracies.
- Automated ticketing systems and Automatic Passenger Counting (APC) systems: This entails comparing the readings of electronic ticketing and Automated Fare Collection (AFC) systems to those of APCs: the difference between the two counting methods directly reports the most common type of fraud: accessing the transit network without validating a ticket. This method is more robust and efficient than manual methods, but it relies heavily on APC accuracy and requires significant investment in technology and infrastructure.
The Promise of AI-Supported Video Analytics
However, there is a new technology that holds promise in addressing the challenge of measuring fare evasion in public transportation: AI-supported video analytics. This system uses machine learning procedures to detect instances of fare evasion in real-time through video analysis at fare gates. AI is much more accurate, consistent and efficient than manual methods, as it can capture all instances of fare evasion individually, regardless of the evasion type. Moreover, it can operate 24/7, providing rapid, complete and detailed data about fare evasion at public transport networks where fare gates or turnstiles control access. The system can also be used to intercept fare evaders in real-time, which has a proven deterring effect.
Conclusion: The Need for More Reliable and Accurate Methods
Overall, the limitations of current fare evasion measurement methods underscore the need for more reliable and accurate methods for measuring fraud in transport. The use of AI-supported video analytics represents a significant advancement in measuring fare evasion in public transportation and has the potential to provide more accurate and reliable data that could be used to improve fare collection and deter fare evasion.
References:
French Ministry for Ecological Transition (2018). “Fraud in transport: diagnosis and recommendations for action”
Egu, O., Bonnel, P. (2020). Can we estimate accurately fare evasion without a survey? Results from a data comparison approach in Lyon using fare collection data, fare inspection data and counting data. Public Transport, Vol. 12, pp. 1–26 E
Juncadella, O. (2019). Article in Intelligent Transport. “Using AI to combat fare evasion”
