Are call detail records biased for sampling human mobility?

Abstract

Call detail records (CDRs) have recently been used in studying different aspects of human mobility. While CDRs provide a means of sampling user locations at large population scales, they may not sample all locations proportionate to the visitation frequency of a user, owing to sparsity in time and space of voice-calls, thereby introducing a bias. Also, as the rate of sampling is inherently dependent on the calling frequencies of an individual, high voice-call activity users are often chosen for conducting a meaningful study. Such a selection process can, inadvertently, lead to a biased view as high frequency callers may not always be representative of an entire population. With the advent of 3G technology and wide adoption of smart-phones, cellular devices have become versatile end-hosts. As the data accessed on these devices does not always require human initiation, it affords us with an unprecedented opportunity to validate the utility of CDRs for studying human mobility. In this work, we investigate various metrics for human mobility studied in literature for over a million cellular users in the San Francisco bay-area, for over a month. Our findings reveal that although the voice-call process does well to sample significant locations, such as home and work, it may in some cases incur biases in capturing the overall spatio-temporal characteristics of individual human mobility. Additionally, we motivate an "artificially" imposed sampling process, vis-a-vis the voice-call process with the same average intensity. We observe that in many cases such an imposed sampling process yields better performance results based on the usual metrics like entropies and marginal distributions used often in literature.

Characterizing and predicting mobile application usage

“…A recent work by [4] showed that call data records can be organized into profiles and clustered using the spatiotemporal usage characteristics of each profile, allowing for accurate prediction and possible adaptation of the network according to the usage dynamics. However, despite these interesting results, there have also been concerns that call data records are biased [14] in the sense that data records are always limited and artificially chosen, and that it may not be a highly accurate method in characterizing general human behavior. Also, mobile user information such as CDR is only so informative as to analyze limited user behavior characteristics such as mobility and crowd information.…”

Data from mobile phone operators: A tool for smarter cities?

“…The findings of such studies can be helpful for policy makers in understanding the characteristics and dynamic nature of different urban areas, as well as updating environmental and (public) transportation policies. Ranjan et al (2012) concluded that the nature of CDRs, which are sparse in time and space, may in some cases cause biases in capturing the overall spatial-temporal characteristics of demographic groups, when studying population level inferences, which may exhibit variable mobility. Finally, Wesolowski et al (2013) have shown that, despite the bias introduced by differential phone ownership, mobility patterns within particular regions are surprisingly robust across populations.…”

“…The meaning of important activity places is analyzed from different statistical approaches and data sets: Paper based travel behaviour data (Flamm and Kaufmann, 2006); WiFi (Kang et al, 2004); GPS (Zhou et al, 2007); GSM in combination with GPS (Nurmi and Koolwaaij, 2006;Nurmi and Bhattacharya, 2008), and CDR's (Ahas, 2006b(Ahas, , 2008bGonzáles et al, 2008;Bayir et al, 2010;Isaacman et al, 2011a;Csáji et al, 2012;Ranjan et al, 2012). Place learning algorithms can be divided into two classes: geometry and fingerprint.…”

Tourism Geography: Emerging Trends and Initiatives to Support Tourism in Morocco

“…For example, in the literature we find an increasing number of empirical studies which use mobile phone data, to identify important activity places (e.g. Ahas et al, 2006aAhas et al, , 2008aAhas et al, , 2010Gonzáles et al, 2008;Kuusik et al, 2008;Bayir et al, 2010;Song et al, 2010a;Phithakkitnukoon et al 2010;Huang et al, 2010;Isaacman et al, 2011a;Csáji et al, 2013;Ranjan et al, 2012;Louail et al, 2014).…”