Mining metro commuting mobility patterns using massive smart card data

With the development of society and economy, the metro has become one of the essential components of the urban transportation system. Commuting passengers prefer the metro due to its punctual, high speeds and uncongested characteristics compared to private cars, taxis, bus, etc., especially in morning and evening rush hour. So, identifying metro commuters and mining its commuting mobility patterns play an essential role in improving service quality, promoting public transit use, and optimizing operational scheduling. We develop a method to mine metro commuting mobility patterns using massive smart card data. Firstly, we extracted individual daily regular OD (origin and destination) based on spatio-temporal similarity measurement from massive smart card data. The information entropy gain algorithm is used to further identify commuters from individual regular OD. Secondly, the station-oriented commute space model is built from space views. Metro stations are divided into employment, residential, and balanced type according to job-housing function pattern. They are divided into high efficiency, general, and low efficiency type according to commute efficiency pattern. Function pattern refers to the proportional relationship between the residence and employment land use around the rail station. Efficiency pattern is a comprehensive index to measure the commute time and distance. Finally, stations are clustered by the K-means method to determine what type they are. The experiment found that metro commuters accounted for 41% of the morning peak traffic using smart card data in Chongqing, China. Three typical job-housing function patterns and three commute efficiency patterns are discovered, respectively, and the characteristics of each are mined.