A Probabilistic Activity Model for Predicting the Mobility Patterns of Homogeneous Social Groups Based on Social Network Data

Francesco Alesiani, Konstantinos Gkiotsalitis, Roberto BaldessariTransportation 2014

Abstract

The prediction of mobility demand is traditionally based on household or individual level travel surveys conducted repeatedly every few years. Due to the unstable economic environment that leads to rapid changes of mobility demand and thanks to the massive amount of real-time data being generated by Smartphones and other ICT devices, complementary models are starting to be considered as potential enablers for a finer and more adaptive travel demand forecasting. In this paper the added value from using user-generated content through social media for mobility demand prediction is examined. A disaggregate model for mobility demand prediction is introduced by linking the mobility decisions of individuals with their activity patterns as they derived by publicly available, user-generated content including geo-tagged information about the location of users. A method for the estimation of individuals activity patterns and another method for classifying individuals into different groups through clustering are also provided. These functions work together to form a mobility prediction model which is tested using data from social media, and its quantization error is presented in two test scenarios. The paper concludes with a remark that social media can be used as a stand-alone or a surrogate source of information for the development of new generation mobility demand prediction models.
BibTeX
@inproceedings{alesiani2014probabilistic,
  author = {Francesco Alesiani and Konstantinos Gkiotsalitis and Roberto Baldessari},
  title = {{A Probabilistic Activity Model for Predicting the Mobility Patterns of Homogeneous Social Groups Based on Social Network Data}},
  booktitle = {Transportation},
  year = {2014},
}

Loading related content...

Figure