How many days are enough?: capturing routine human mobility. Issue 7 (3rd July 2018)

• Record Type: Journal Article
• Authors: Stanley, Kevin; Yoo, Eun-Hye; Paul, Tuhin; Bell, Scott
• Is Part Of: International journal of geographical information science. Volume 32:Issue 7(2018)
• Page Start: 1485
• Page End: 1504
• Publication Date: 2018-07-03
• Subjects: Activity spaces -- Kullback–Leibler (KL) divergence -- Global Positioning Systems (GPS); Geography -- Data processing -- Periodicals; Information storage and retrieval systems -- Periodicals; Géomatique -- Périodiques; Systèmes d'information -- Périodiques; 910.285
• DOI: 10.1080/13658816.2018.1434888
• Languages: English
• ISSNs: 1365-8816

Abstract

ABSTRACT: Wedding mobile phone sensor technology and human spatial behaviour has great potential. The ubiquity of Global Positioning Systems (GPS) technology has made gathering data about human mobility simpler, more precise, and with higher fidelity, providing minute-by-minute records of the locations of cohorts from dozens of participants. While this data provides a strong basis for Geographic Information Science research, it also constitutes an invasion of the participants' privacy and can provide more information than researchers require to answer their questions. As an ethical and practical consideration, researchers should gather only as much data as they need. In this paper, we take three weeks of GPS traces from over a hundred student participants in mobile phone-based tracking studies and show that fewer than 14 days of data is necessary to establish complete activity spaces. We define 'complete' as the point at which marginal information gains become negligible according to a pairwise temporal analysis of the Kullback–Leibler (KL) divergence of the spatial (bivariate) histogram through time. For the fixed level of information difference, observable in the data, impacts due to individual variability, population composition, and spatial resolution are evident. However, all populations at each level of resolution examined in the paper demonstrated convergence to low divergence levels occurred within a matter of days, and to negligible information gain in less than two weeks. The methods described in the paper represent a novel metric useful to understand the interaction between measurements and information in human mobility.