Detecting Pattern Changes in Individual Travel Behavior from Vehicle GPS/GNSS Data

Although stable in the short term, individual travel behavior generally tends to change over the long term. The ability to detect such changes is important for product and service providers in continuously changing environments. The aim of this paper is to develop a methodology that detects changes in the patterns of individual travel behavior from vehicle global positioning system (GPS)/global navigation satellite system (GNSS) data. For this purpose, we first define individual travel behavior patterns in two dimensions: a spatial pattern and a frequency pattern. Then, we develop a method that can detect such patterns from GPS/GNSS data using a clustering algorithm. Finally, we define three basic pattern-change scenarios for individual travel behavior and introduce a pattern-matching metric for detecting these changes. The proposed methodology is tested using GPS datasets from three randomly selected anonymous users, collected by a Chinese automotive manufacturer. The results show that our methodology can successfully identify significant changes in individual travel behavior patterns.