Whereabouts of truckers: An empirical study of predictability

Mi Gan, Yu (Marco) Nie*, Xiaobo Liu, Daoli Zhu

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


This study examines the predictability of a truckload trucker's whereabouts, using a GPS trajectory data set from about 1000 active users of an on-line freight exchange (OFEX) platform. To describe a trucker's whereabouts, two different location mapping schemes are proposed. The first divides the entire study area into rectangular grids using a simple geographic rescaling method and treats each grid as a unique location. In the second scheme, a location type (LT) is first assigned to each GPS point, using Point Of Interest (POI) information. Then, each GPS trajectory is converted to a trajectory of LT data. We then process both types of trajectory data, grid-based and type-based, to develop various entropy and predictability measures. We find that the whereabouts of truckers possess strong regularity, but are substantially more difficult to predict than that of an average person. Specifically, the predictability of truckers’ next location on the grid map peaks at 83%, about 10% percentage point lower than that reported in the literature. Surprisingly, predicting the type of the location that a typical trucker may visit the next is even more difficult, with a predictability around 81%. Findings from this study could help guide the development of algorithms for predicting truckers’ next visit location and the type of the location.

Original languageEnglish (US)
Pages (from-to)184-195
Number of pages12
JournalTransportation Research Part C: Emerging Technologies
StatePublished - Jul 2019


  • Entropy
  • On-line freight exchange
  • Predictability
  • Regularity
  • Whereabouts

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Automotive Engineering
  • Transportation
  • Computer Science Applications


Dive into the research topics of 'Whereabouts of truckers: An empirical study of predictability'. Together they form a unique fingerprint.

Cite this