User:Deeplumi/sandbox/Human Mobility Patterns


Patterns of Human Mobility

Individual human mobility is the study of how individual humans move within given geographical and temporal boundaries.[1] In this context, the patterns of human mobility are the rules that govern the motion of individuals aggregated over some spatio-temporal resolutions. Research in human mobility is an interdisciplinary field, which draws heavily on methods from social, natural and computational sciences.

Since the temporal and spatial dimensions of human mobility are often represented as a complex network system, research in this field is considered to be an integral part of network science domain.

Research history

edit

The concept of human mobility is deeply rooted in social science research tradition. The first systematic analysis of the concept of distance as a constraint to movement is often attributed to Henry C. Carey’s Principles of Social Science (1858), which was the first to explicitly make the observation about the amount of interaction between two cities being proportional to their population size and inversely proportional to the intervening distance.

Few decades later, the geographer Ernst Ravenstein further developed and popularized the idea in a seminal work where he formulated his laws of migration[2]. Further refinements on this theme were made in the 1940’s by the sociologist Samuel Stouffer[3] in his law of intervening opportunities, and by the economist William J. Reilly, who formulated the gravity model of migration[4].

The increasing availability of datasets on population movements at various levels of granularity, coupled with the quantitative revolution and development of computational methods, led to the introduction of more sophisticated mathematical methods such as hidden Markov models and diffusion processes. As a result, towards the turn of the XXI century the research on human mobility transcended its initial boundaries of social sciences and human geography and emerged as a separate interdisciplinary field of study.

Data and methods

edit

Historically, information on individual mobility was extracted from travel surveys, which provided the detailed itinerary of individual travel. They were designed to gather rich information about travel choices but were expensive to distribute over the entire population, covering typically, only a very limited time span and number of individuals.

Recently, the rich amount of information containing individual’s coordinates is automatically recorded each time a person makes a phone call, sends an email, uses a credit card, or travels using a public transport smart card etc became available. This traces of coordinates are recorded over months or even years. These data sources offered researchers the unique opportunity to understand and characterize the patterns of human travel behavior at a massive scale, not feasible before.

However, these data come with their own limitations. The sources often contain incomplete information about individuals trips; which brings particular challenges for relating individual mobility to conventionally used models. Quite often, these data sources lack the demographic information of individuals, the traces of their locations are not continuously collected and there is no detailed information about the specific choices of the individual's daily activities.

From the methodology standpoint, extracting human mobility patterns involves systematic characterization of aggregated individual’s longitudinal mobility patterns from alternative data sources, possibly separately for groups of individuals that are distinct in their behavior. Recent research has shown that patterns of human mobility, which can be extracted across spatio-temporal domains, depend on the properties of travel itself (i.e., long-distance vs intra-urban travel), but also on the personal habits of the moving individuals.

The modeling of human mobility and extracting its characteristic patterns had adopted new directions due to the increasing availability of big data sources from human activity. Despite this recent availability of large scale data on human mobility, a full understanding of the rules governing motion within social systems is still missing, mostly due to incomplete information on the socio-economic factors and to often limited spatio-temporal resolutions. However, since the beginning of 2010s a significant progress was made in this area mainly due to the employment of new models and computational approaches, particularly neural networks and deep learning algorithms.

Statistical properties of human mobility

edit

Distribution of trajectories

edit

Human mobility studies, from a statistical physics perspective, began with analyzing animal trajectories, where the trajectories were approximated by scale-free random walks known as Lévy flights[5]. Extending these results to human mobility, the distribution of traveling distances on a high spatial scale (between 10–3200 km) decays as a power law suggesting that human trajectories are best modeled as a continuous-time random walk [25].

Pioneer studies, based on call detail record (CDR)[6] and banknote records[7], found that the distribution of displacement ( ) is well approximated by a power-law:  , or Levy distribution, as typically  , and that an exponential cut-off in the distribution may control boundary effects [2]. This indicates that individual human mobility is very heterogenous in its nature.

Recent research however found , that power law distributions are selected as the best model only when large spatial or temporal scales are considered. For the smaller spatial and temporal scales it was shown that log-normal distributions characterize the distribution of displacements, implying that this property is not a simple consequence of the stability of human mobility, but also a characteristic feature of human behavior. Another analysis have shown that the short trip distribution, that is mainly related to urban mobility, does not follows a power law, but closely obeys the exponential law. Additionally, the study of the downtime distribution between two successive trips has confirmed the existence of a universal Benford’s law that could be related to a log-time perception, when individuals perform their daily ”asystematic” activities[8].

Variability and predictability

edit

A relatively counterintuitive finding of research in individual mobility patterns is that high variability in characteristic travel patterns of individuals (e.g. distances) coexists with high predictability of future travel locations. In other words, while individuals differ greatly in their travel distances, downtimes and visited locations, the individual trajectories can be predicted with amazingly high degree of certainty (over 90%).[9]

Preferential return

edit

An important aspect of human behavior is the tendency of individuals to return to one or several locations on a daily basis (so called preferential return). This reflects the fact the most individuals tend to travel between several, generally 3 to 5, locations, such as home, workplace or locations of leisure activities (gym, movie theater, restaurant etc). Such set of predominantly visited locations combined with the temporal patterns of travel between them is often referred to as routine travel pattern of a particular individual. This routine can be temporally broken down, for example by going on vacation, but as research has shown, in general, the pattern is stable over aggregated time period for most individuals[10].

Long-distance and intra-urban mobility

edit

It is an established fact that large scale mobility patterns follow the power law. However, considering that most individuals’ movement is limited within particular urban areas, the research of intra-urban mobility patterns became quite prominent aspect of the overall research in human mobility. Results indicate that the distribution of human’s intra-urban travel in general follows the exponential law. The exponents, however, vary from city to city and indicate the impact of city sizes and shapes. Individuals living in large or less compact cities generally need to travel farther on a daily basis, and vice versa[11].

Daily mobility patterns

edit

The research has shown that almost universally in respect to temporal (e.g., time of year) and geographical coordinates, three peaks are normally detected in the aggregated daily mobility patterns correspond to the systematic activities of jobs-related activities, leisure and the night rest. Analysis points out that the majorities of the performed activities seems to obey to a statistical distribution of independent events even if the time spent in the ”systematic activities” is relevant. This may indicate that human behavior during mobility has strong stochastic components and the predictability of the spatial displacements can be difficult, contrarily to the localization in time. A study aimed to investigate the relevance of daily habits in the individual mobility[12], confirmed the idea that the individual mobility patterns can be understood using a preferential attachment paradigm, but the comprehension of the underlying mechanisms of the individual mobility demand is still an open problem.

Social ties based mobility

edit

It is natural to assume that two individuals who have a social interaction, such as friends, family or colleagues, do not always move independently. With some regularity, they will share full trips, destinations or origins. The trips can be also synchronized if the objective is to meet somewhere or go back home after a meeting. Furthermore and closing the loop, the social network of an individual typically reflects the geography of their life with tighter connections with people spatially closer or at least in clusters related to the places in which the person has previously resided. These correlations have been observed in several publications.[13] Additionally, the relation between distance and online friendships was confirmed using surveys, social networks and mobile phone records.

Individual traveling behavior

edit

As was noted by many researchers, there are some properties of individual human mobility which do not stem from the regular statistical patterns, but depend on the individual traveling behavior. For example, a study[14] has shown that based on the proportion of recurrent mobility - i.e., traveling between n most often visited locations - to the overall mobility, the population can be divided into explorers and returners. Explorers and returners have distinct mobility patterns. Returners limit much of their mobility to a few locations, hence their recurrent and overall mobility are comparable. In contrast, the mobility of explorers cannot be reduced to few locations, they tend to explore more. Interestingly, it was noted that individuals tend to engage in social interactions with individuals with similar mobility profiles. However, while the existence of a strong correlation between the mobility behavior of individuals and their social relationships is observed, further research is needed to understand whether this can be interpreted as a homophily or influence effect.

Applications

edit

Models of human mobility have a wide range of applications. Some of those can be direct approaches such as traffic forecasting, activity based modeling, and design of transportation networks, while others may require more specific models of human mobility such as the spreading of viruses and the evolution of epidemics, measuring human exposure to air pollutants, and simulating mobile networks of wireless devices. Analysis of traveling behavior of individuals and their groups can also have applications for business decisions, such as targeted marketing of products, advertising and promotional campaigns.


References

edit
  1. ^ Keyfitz, Nathan (1973). "Individual Mobility in a Stationary Population". Population Studies. 27 (July 1, 1973): 335–352. doi:10.2307/2173401. JSTOR 2173401.
  2. ^ Ravenstein, E. G. (1885-06-01). "The Laws of Migration". Journal of the Statistical Society of London. 48 (2): 167–235. doi:10.2307/2979181. ISSN 0959-5341. JSTOR 2979181.
  3. ^ Stouffer, Samuel A. (1940-12-01). "Intervening Opportunities: A Theory Relating Mobility and Distance". American Sociological Review. 5 (6): 845–867. doi:10.2307/2084520. ISSN 0003-1224. JSTOR 2084520.
  4. ^ John., Reilly, William (1959). Methods for the study of retail relationships. Univ. of Texas, Bureau of Business Research. OCLC 221187930.{{cite book}}: CS1 maint: multiple names: authors list (link)
  5. ^ Sims, David W.; Southall, Emily J.; Wearmouth, Victoria J.; Noble, Leslie R.; Jones, Catherine S.; Hays, Graeme C.; Houghton, Jonathan D. R.; Doyle, Thomas K.; Brunnschweiler, Juerg M. (2010-06-01). "Environmental context explains Lévy and Brownian movement patterns of marine predators". Nature. 465 (7301): 1066–1069. Bibcode:2010Natur.465.1066H. doi:10.1038/nature09116. ISSN 1476-4687. PMID 20531470.
  6. ^ Barabási, Albert-László; Hidalgo, César A.; González, Marta C. (2008-06-01). "Understanding individual human mobility patterns". Nature. 453 (7196): 779–782. arXiv:0806.1256. Bibcode:2008Natur.453..779G. doi:10.1038/nature06958. ISSN 1476-4687. PMID 18528393.
  7. ^ Geisel, T.; Hufnagel, L.; Brockmann, D. (2006-01-01). "The scaling laws of human travel". Nature. 439 (7075): 462–465. arXiv:cond-mat/0605511. Bibcode:2006Natur.439..462B. doi:10.1038/nature04292. ISSN 1476-4687. PMID 16437114.
  8. ^ Barabási, Albert-László; Blumm, Nicholas; Qu, Zehui; Song, Chaoming (2010-02-19). "Limits of Predictability in Human Mobility". Science. 327 (5968): 1018–1021. Bibcode:2010Sci...327.1018S. doi:10.1126/science.1177170. ISSN 1095-9203. PMID 20167789.
  9. ^ Barabási, Albert-László; Blumm, Nicholas; Qu, Zehui; Song, Chaoming (2010-02-19). "Limits of Predictability in Human Mobility". Science. 327 (5968): 1018–1021. Bibcode:2010Sci...327.1018S. doi:10.1126/science.1177170. ISSN 1095-9203. PMID 20167789.
  10. ^ Pappalardo, Luca; Simini, Filippo (2017-12-27). "Data-driven generation of spatio-temporal routines in human mobility". Data Mining and Knowledge Discovery. 32 (3): 787–829. doi:10.1007/s10618-017-0548-4. ISSN 1384-5810.
  11. ^ Kang, Chaogui; Ma, Xiujun; Tong, Daoqin; Liu, Yu (2012-02-15). "Intra-urban human mobility patterns: An urban morphology perspective". Physica A: Statistical Mechanics and its Applications. 391 (4): 1702–1717. Bibcode:2012PhyA..391.1702K. doi:10.1016/j.physa.2011.11.005. ISSN 0378-4371.
  12. ^ Riccardo, Gallotti; Armando, Bazzani; Sandro, Rambaldi (2012-07-23). "Towards a statistical physics of human mobility". International Journal of Modern Physics C. 23 (9): 1250061. arXiv:1207.5698. Bibcode:2012IJMPC..2350061R. doi:10.1142/S0129183112500611. ISSN 0129-1831.
  13. ^ Baronchelli, Andrea; Lehmann, Sune; Sapiezynski, Piotr; Alessandretti, Laura (2017-02-15). "Multi-scale spatio-temporal analysis of human mobility". PLOS ONE. 12 (2): e0171686. arXiv:1609.05514. Bibcode:2017PLoSO..1271686A. doi:10.1371/journal.pone.0171686. ISSN 1932-6203. PMC 5310761. PMID 28199347.{{cite journal}}: CS1 maint: unflagged free DOI (link)
  14. ^ Barabási, Albert-László; Giannotti, Fosca; Pedreschi, Dino; Rinzivillo, Salvatore; Simini, Filippo; Pappalardo, Luca (2015-09-08). "Returners and explorers dichotomy in human mobility". Nature Communications. 6: 8166. Bibcode:2015NatCo...6E8166P. doi:10.1038/ncomms9166. ISSN 2041-1723. PMC 4569739. PMID 26349016.