Uncovering Measurements of Social and Demographic Behavior From Smartphone Location Data

Daniel Kelly, Smyth Barry, Caulfield Brian

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

Human behavior, and in particular location behavior, is highly routine based. Modern mobile phones, through global position system (GPS) technology and cell tower and WiFi location identification, enable us to trace human location behavior at scales that were previously unattainable. The goal of this paper is to examine human location behavior, through mobile phone data, and investigate if links can be made between location behavior patterns and particular demographic and social characteristics about an individual. We hypothesize that an individual's daily predictability can be key to linking their behavior to certain characteristics, and we propose predictability and geographic areas of interest models to analyze this hypothesis. Experiments reveal that measurements, which are based on our proposed location predictability models, can correctly infer 17 different characteristics about an individual with an average accuracy of 85.5%.
LanguageEnglish
Pages188-198
JournalIEEE Transactions Human-Machine Systems
Volume43
Issue number2
DOIs
Publication statusPublished - 31 Mar 2013

Fingerprint

Smartphones
Mobile phones
Towers
Experiments

Keywords

  • Human Behavior
  • GPS
  • Smartphone
  • Clustering
  • Hidden Markov Model

Cite this

@article{00689e55fc85428aabb3c0b1e084c46b,
title = "Uncovering Measurements of Social and Demographic Behavior From Smartphone Location Data",
abstract = "Human behavior, and in particular location behavior, is highly routine based. Modern mobile phones, through global position system (GPS) technology and cell tower and WiFi location identification, enable us to trace human location behavior at scales that were previously unattainable. The goal of this paper is to examine human location behavior, through mobile phone data, and investigate if links can be made between location behavior patterns and particular demographic and social characteristics about an individual. We hypothesize that an individual's daily predictability can be key to linking their behavior to certain characteristics, and we propose predictability and geographic areas of interest models to analyze this hypothesis. Experiments reveal that measurements, which are based on our proposed location predictability models, can correctly infer 17 different characteristics about an individual with an average accuracy of 85.5{\%}.",
keywords = "Human Behavior, GPS, Smartphone, Clustering, Hidden Markov Model",
author = "Daniel Kelly and Smyth Barry and Caulfield Brian",
year = "2013",
month = "3",
day = "31",
doi = "10.1109/TSMC.2013.2238926",
language = "English",
volume = "43",
pages = "188--198",
journal = "IEEE Transactions on Human-Machine Systems",
issn = "2168-2291",
number = "2",

}

Uncovering Measurements of Social and Demographic Behavior From Smartphone Location Data. / Kelly, Daniel; Barry, Smyth; Brian, Caulfield.

In: IEEE Transactions Human-Machine Systems, Vol. 43, No. 2, 31.03.2013, p. 188-198.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Uncovering Measurements of Social and Demographic Behavior From Smartphone Location Data

AU - Kelly, Daniel

AU - Barry, Smyth

AU - Brian, Caulfield

PY - 2013/3/31

Y1 - 2013/3/31

N2 - Human behavior, and in particular location behavior, is highly routine based. Modern mobile phones, through global position system (GPS) technology and cell tower and WiFi location identification, enable us to trace human location behavior at scales that were previously unattainable. The goal of this paper is to examine human location behavior, through mobile phone data, and investigate if links can be made between location behavior patterns and particular demographic and social characteristics about an individual. We hypothesize that an individual's daily predictability can be key to linking their behavior to certain characteristics, and we propose predictability and geographic areas of interest models to analyze this hypothesis. Experiments reveal that measurements, which are based on our proposed location predictability models, can correctly infer 17 different characteristics about an individual with an average accuracy of 85.5%.

AB - Human behavior, and in particular location behavior, is highly routine based. Modern mobile phones, through global position system (GPS) technology and cell tower and WiFi location identification, enable us to trace human location behavior at scales that were previously unattainable. The goal of this paper is to examine human location behavior, through mobile phone data, and investigate if links can be made between location behavior patterns and particular demographic and social characteristics about an individual. We hypothesize that an individual's daily predictability can be key to linking their behavior to certain characteristics, and we propose predictability and geographic areas of interest models to analyze this hypothesis. Experiments reveal that measurements, which are based on our proposed location predictability models, can correctly infer 17 different characteristics about an individual with an average accuracy of 85.5%.

KW - Human Behavior

KW - GPS

KW - Smartphone

KW - Clustering

KW - Hidden Markov Model

U2 - 10.1109/TSMC.2013.2238926

DO - 10.1109/TSMC.2013.2238926

M3 - Article

VL - 43

SP - 188

EP - 198

JO - IEEE Transactions on Human-Machine Systems

T2 - IEEE Transactions on Human-Machine Systems

JF - IEEE Transactions on Human-Machine Systems

SN - 2168-2291

IS - 2

ER -