Moving and calling: Mobile phone data quality measurements and spatiotemporal uncertainty in human mobility studies

Abstract

In the past few years, mobile network data are considered as a useful complementary source of information for human mobility research. Mobile phone datasets contain massive amount of spatiotemporal localization of millions of users. The analyze of such huge amount of data for mobility studies reveals many issues such as time computation, users sampling, spatiotemporal heterogeneities, semantic incompleteness. In this chapter, two issues are addressed: (1) location sampling aiming at decreasing computation time without losing useful information on the one hand and to eliminate data considered as noise in the other hand and (2) users sampling whose goal is to select users having relevant information. For the first issue two measures allowing eliminating redundant information and pingpong positions are proposed. The second issue requires the definition of a set of measures allowing estimating mobile phone data quality. New methods to qualify mobile phone data at local and global level are proposed. The methods are tested on one-day mobile phone data coming from technical mobile network probes.