| Application: |
Clustering Smart Card Data for Urban Mobility Analysis |
| City: |
Rennes, France |
| Organization: |
This work was supported by the Research and Innovation Program in land transport (PREDIT). |
| Project Start Date: |
Unknown. |
| Project End Date: |
3 March 2017 (Published) |
| Reference: |
El Mahrsi, Mohamed K and Come, Etienne and Oukhellou, Latifa and Verleysen, Michel (2017). Clustering Smart Card Data for Urban Mobility Analysis. IEEE Transactions on Intelligent Transportation Systems, ISSN 1524-9050, 2017, Volume 18, Issue 3, pp. 712 - 728 |
| Problem: |
Smart card data present a unique opportunity to study passenger travel behavior in public transportation systems. This study was done in Rennes France is attempting to analyze the data collected from automated fare collection (AFC) and ultimately to be used for transit planning. |
| Technical Solution: |
Two approaches are considered for clustering the unsupervised data collected which can be used to extract mobility patterns in a public transportation system:
- A station-oriented operational point of view: Clusters stations based on when their activity occurs, i.e., how trips made at the stations are distributed over time.
- Passenger-focused one: Identify groups of passengers that have similar boarding times aggregated into weekly profiles.
|
| Datasets Used: |
Data collected through the automated fare collection system of the Service des Transports en commun de l’Agglomération Rennaise (STAR). STAR operates over 70 regular bus lines (excluding school bus and complementary services) and 1 subway line serving the metropolitan area of Rennes, France.
The original dataset spans over a one-month period (April 2014) and contains a total of 5404096 journey transactions out of which 4325839 (80% of the data) were made by 134979 smart cards.
|
| Outcome: |
For stations Clustering the results can be broken into 2 categories :
- Balanced usage with several peaks during rush hour.
- Unbalanced usage in which the number of transactions during one half of the day
For stations Clustering the results:
- 13 passenger clusters are discovered.
- Clusters are rather characterized by a diffuse usage of public transit that appears through different times of the day.
|
| Issues that arose: |
Using smart card data to analyze human mobility raises challenges due to :
- Big volume: Large amount of transactions are registered per day.
- Incompleteness: Trip destination information (particularly for trips involving multiple stages) are often missing. Data about trip purposes are also unavailable. These information play a key role not only in understanding travel behavior
- Lack of socioeconomic data aimed to protect passenger privacy; socioeconomic indicators (age, gender, revenue, etc.) are omitted, despite their usefulness in conducting detailed travel behavior analyses.
|
| Status: |
Results were promising and future studies will be conducted to explore more data. |
| Entered by: |
Yaser Khalil, yaser.khalil@mail.utoronto.ca |