From a transit authority perspective, it is important to understand how users move in the network. The paper proposes using fare cohorts as a unit of analysis in order to examine understand market segments and review fare policy.

We use internal information from the smart card fare validation data to build fare cohorts. We found that trip frequency is the most representative characteristic. We looked at 8 days of trip history and categorized almost 1.5 million cards by trip frequency. We have 3 groups excluding single-use fare media: frequent users, occasional users and those who travelled only once. We can see distinctive patterns of fare media, fare zone and fare groups. We could have tried clustering to create groups.

Here is the data flow. It has been built on many years of research and constant data collection and maintenance. We merge individual fare validation data with operations and geospatial data to get vehicle location and route geometry. We then assign each fare transaction a fare cohort. We derive the alighting stop to reconstruct the complete trip OD, along with a trip purpose with an algorithm trained with regional OD survey. At last, we assign a real-world location for each origin and destination, a socio-demographic profile for each ID. We end up with a file that describe more than 95% of the observed demand.

Overall, we have applied data merging and enrichment techniques with massive amount of data. The resulting trip file would allow transit authority to answer many questions on fare policy and network consumption, such as adjusting level of service, refining fare level, proposing more equitable fare structure and funding mechanism and initiatives that aim to improve customer experience of targeted segments.