Abstract:
As people become more mobile, urban traffic patterns become more complex, creating a need for more continuous transportation planning processes. Currently, manual and online surveys are the primary source for such analysis. However, such data collection while being expensive, takes time and is often outdated by the time it is made available for analysis. Mobile Network Big Data (MNBD) which concerns large data sets has the potential to supplement such traditional data sampling programs. Call Detail Records (CDR) which is a subset of MNBD is readily available as most of the telecommunication service providers maintain such data. Thus, analyzing CDR leads to an efficient identification of human behavior and location.
This research uses the CDRs of nearly 10,000 mobile phone users in the Western Province (WP) of Sri Lanka for a period of three months for the analysis of their caller locations in order to determine their mobility patterns. In analyzing the CDRs, the frequency of making calls from a specific location is identified, classified them into potential home and non-home locations based on the regularity and time of day and week these calls were generated from each such location. Users are thereby categorized hierarchical levels based on the regularity of presence within the study area, identification of province of residence and by typical employment categories across the sample of users.
An estimate of home-based work trips made within the Western Province was identified using the CDR and validated by comparing with the origin-destination matrix for work trips calculated from an extensive survey of 35,000 households under the CoMTrans Study (JICA, 2014) obtaining a fit of 76%. The successful validation and the identification of sources of errors in CDR data provides direction for further research in using CDRs for travel estimation and the identification of the appropriate comprehensive data mining techniques.