Public Transport Smart Card Data Analysis Using Tucker Decomposition
With the development of information and communication technology, various kinds of data have been collected and accumulated in large quantities. Along with this, many methods for analyzing large-scale data have been proposed. This study extracts travel characteristics from smart card data of a private rail system using Tucker decomposition, a valid method for analyzing high order data. Applying Tucker decomposition to a 6th-order tensor dataset (weatherday of the weektime of daypassenger typeorigin stationdestination station), the results reveal what kind of users move from which station to which station, in what weather, on which days of the week, and at what time of day. Moreover, the dataset of 6,489,600 elements was compressed to 1,440 elements with the suggested approach, and the main travel patterns of passengers could be confirmed.
KeywordsTensor factorization Smart card High order data Travel pattern
The authors especially thank Takamastu-Kotohira Electric Railroad Co. in Takamatsu City, Kagawa Prefecture, Japan for providing smart card data.
- 17.Vazifehdan, M., Moattar, M.H., Jalali, M.: A hybrid Bayesian network and tensor factorization approach for missing value imputation to improve breast cancer recurrence prediction. J. King Saud Univ. Comput. Inf. Sci. 31, 175–184 (2018)Google Scholar
- 21.Han, Y., Moutarde, F.: Analysis of large-scale traffic dynamics in an urban transportation network using non-negative tensor factorization. Int. J. Intell. Transp. Syst. Res. 14, 36–49 (2014)Google Scholar