Understanding User Attributes from Calling Behavior
Open Access
- 8 December 2014
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 95-104
Abstract
Mobile phones are arguably one of the most prolific sources of large-scale human mobility data. The availability of this data has generated a massive body of research focused on understanding the dynamics and patterns of human mobility. However, it is increasingly evident that additional value can be derived from such data. This paper proposes a novel approach for understanding the attributes of mobile users by analyzing calling behavior derived from field survey data, in combination with call detail records (CDRs). Our survey reveals distinctive traits in calling behavior that correspond to user attributes. Analysis results demonstrate that frequent call locations, the variability in call time distributions, and the locations from which calls are made around midday are all keys to distinguishing gender. In addition, the location of calls initiated during the morning hours is a key to analyzing income levels for males.Keywords
This publication has 16 references indexed in Scilit:
- Human mobility modeling at metropolitan scales2012
- Predictability of population displacement after the 2010 Haiti earthquakeProceedings of the National Academy of Sciences, 2012
- Predicting Customer Demographics in a Mobile Social Network2011
- Identifying Important Places in People’s Lives from Cellular Network Data2011
- Mobile divides2010
- Mining user similarity based on location history2008
- Understanding individual human mobility patternsNature, 2008
- Learning the parts of objects by non-negative matrix factorizationNature, 1999
- Socio-demographics, activity participation and travel behaviorTransportation Research Part A: Policy and Practice, 1999
- Least squares formulation of robust non-negative factor analysisChemometrics and Intelligent Laboratory Systems, 1997