Créer des clusters de trajectoires d'acheteurs par les structures de réseau

Translated title of the contribution: Clustering customer shopping trips with network structure

Ping Yan, Daniel D. Zeng

Research output: Contribution to conferencePaperpeer-review

3 Scopus citations


Moving objects can be tracked with sensors such as RFID tags or GPS devices. Their movement can be represented as sequences of time-stamped locations. Studying such spatio-temporal movement sequences to discover spatial sequential patterns holds promises in many real-world settings. A few interesting applications are customer shopping traverse pattern discovery, vehicle traveling pattern discovery, and route prediction. Traditional spatial data mining algorithms suitable for the Euclidean space are not directly applicable in these settings. We propose a new algorithm to cluster movement paths such as shopping trips for pattern discovery. In our work, we represent the spatio-temporal series as sequences of discrete locations following a pre-defined network. We incorporate a modified version of the Longest Common Subsequence (LCS) algorithm with the network structure to measure the similarity of movement paths. With such spatial networks we implicitly address the existence of spatial obstructs as well. Experiments were performed on both hand-collected real-life trips and simulated trips in grocery shopping. The initial evaluation results show that our proposed approach, called Net-LCSS, can be used to support effective and efficient clustering for shopping trip pattern discovery.

Translated title of the contributionClustering customer shopping trips with network structure
Original languageFrench
StatePublished - Dec 1 2008
Event29th International Conference on Information Systems, ICIS 2008 - Paris, France
Duration: Dec 14 2008Dec 17 2008




  • Path clustering algorithms
  • RFID application
  • Shopping behavior study

ASJC Scopus subject areas

  • Information Systems


Dive into the research topics of 'Clustering customer shopping trips with network structure'. Together they form a unique fingerprint.

Cite this