Abstract
Understanding traffic conditions in urban areas is an important research direction for efficient transportation management, and demand for detailed and immediate information is increasing. User-anonymized mobile phone billing records are now known to have an especially high potential for effective traffic conditions estimation, due to their wide population and area coverage. The purpose of this paper is to propose a method to estimate full day mobility of mobile phone users based on long term Call Detail Records (CDR) data. There are two main features of this method. One is that the estimated route taken by a candidate reflects past mobility patterns extracted from past CDR records. Another is that the method does not rely on the spatio-temporal resolution of CDR data, and is applicable to sparse datasets. The proposed method is tested with actual CDR dataset with 4 different temporal resolution to see the accuracy in estimated mobility mode and location change. The estimated results showed more than 72% accuracy in mobility mode, and less than 1.5km location difference with GPS location information.
