Source-linked AI summary
A city of cities: Measuring how 15-minutes urban accessibility shapes human mobility in Barcelona
Eduardo Graells-Garrido, Feliu Serra-Burriel, Francisco Rowe, Fernando M. Cucchietti, Patricio Reyes
TL;DR
Evidence is limited on how local urban accessibility relates to human mobility across neighborhoods, despite the relevance of local amenities to 15-minute-city planning. Using mobile phone, census, and volunteered geographical data in a Negative Binomial Geographically Weighted Regression, the paper finds wide geographic variability in these relationships.
Problem
How local amenity availability relates to human mobility across neighborhoods remains little understood, despite its relevance to polycentric urban planning.
Source: Introduction · pp. 2–3Method
The study combines mobile phone, census, and volunteered geographical data with a Negative Binomial Geographically Weighted Regression to examine mobility and local accessibility in Barcelona.
Source: Materials and methods · p. 4Results
The relationship between human mobility and local amenity accessibility varies widely across neighborhoods, with local amenity associations changing in sign and magnitude.
Source: Discussion and Conclusions · pp. 22–23Takeaways & Limitations
The findings contribute to informing spatially targeted urban policy.
Source: Discussion and Conclusions · p. 23The analysis does not model temporal variation, although methods extending GWR with temporal awareness are emerging.
Source: Discussion and Conclusions · p. 25
Abstract
from arXiv · showhide
As cities expand, human mobility has become a central focus of urban planning and policy making to make cities more inclusive and sustainable. Initiatives such as the "15-minutes city" have been put in place to shift the attention from monocentric city configurations to polycentric structures, increasing the availability and diversity of local urban amenities. Ultimately they expect to increase local walkability and increase mobility within residential areas. While we know how urban amenities influence human mobility at the city level, little is known about spatial variations in this relationship. Here, we use mobile phone, census, and volunteered geographical data to measure geographic variations in the relationship between origin-destination flows and local urban accessibility in Barcelona. Using a Negative Binomial Geographically Weighted Regression model, we show that, globally, people tend to visit neighborhoods with better access to education and retail. Locally, these and other features change in sign and magnitude through the different neighborhoods of the city in ways that are not explained by administrative boundaries, and that provide deeper insights regarding urban characteristics such as rental prices. In conclusion, our work suggests that the qualities of a 15-minutes city can be measured at scale, delivering actionable insights on the polycentric structure of cities, and how people use and access this structure.
1 Barcelona Supercomputing Center (BSC), Barcelona, Catalonia, Spain · 4 Data Science Institute, Universidad del Desarrollo, Santiago, Chile · Introduction
The introduction frames the study around the gap between polycentric, 15-minute-city planning goals and limited evidence on how local amenities shape mobility across neighborhoods. It proposes combining mobile-phone, census, and OpenStreetMap data with global and geographically weighted regression to measure those relationships in Barcelona.
Introduction
- Urban sprawl is linked to longer trips, car use, congestion, obesity, and pollution, while commute lengths also differ across population groups.The introduction notes longer commutes among highly educated, affluent, male, and dual-earner populations, with different patterns in less developed and unequal societies.Source: Introduction · p. 2
- Planning strategies promote localized mobility by using smart-growth interventions to move from monocentric structures toward polycentric configurations, while 15-minute-city ideas gain policy salience and interventions increase local amenities.Source: Introduction · p. 2
- Prior work links concentrated jobs, public services, and commerce to mobility hotspots, but evidence is limited on local amenity retention and geographic variation by resident demographics.The study addresses how the relationship between local amenities and mobility intensity varies geographically and interacts with the socio-demographic profile of residents.Source: Introduction · p. 3
- The study models Barcelona residents’ origin-destination flows using OSM amenity data, census demographics, Negative Binomial regression, and Negative Binomial GWR.The global model identifies system-wide amenities associated with low mobility intensity or high retention and examines population-group differences; GWR measures the direction and extent of spatial variation.Source: Introduction · pp. 3–4
- Surveys and travel diaries have limited frequency, coverage, and fine-scale statistical validity, whereas anonymized mobile-phone data provide contemporary mobility detail at high temporal and geographic granularity [26–29].The introduction notes that surveys may appear one or two years after collection, while phone data represent selected users and raise privacy and confidentiality concerns.Source: Introduction · p. 3
- The analysis produces neighborhood-level accessibility and mobility vectors that can differ from citywide patterns and supports further local analysis of their relationship with rental prices.Barcelona’s aggregated mobile-phone data were accessed through a collaboration between the city council and Vodafone.Source: Introduction · pp. 3–4
Materials and methods
The methods combine census data, mobile-phone flows, and walking-accessibility measures to characterize neighborhood mobility and amenity access, model origin–destination flows at citywide and local scales, and examine how local mobility patterns relate to rental attributes.
Materials and methods
- The methods combine neighborhood accessibility measures with census and mobile-phone data to identify local patterns, then use those patterns to explore mobility–rental-price relationships.Source: Materials and methods · p. 4
City data
- Barcelona’s 73 neighborhoods have 2018 census measures including socioeconomic and demographic characteristics and rental prices; their spatial heterogeneity supports a local approach to mobility.Source: City data · p. 5
Aggregated mobile phone data flows
- The analysis focuses on residents’ destination visits, using June mobile-phone counts and converting a 212-cell destination grid into neighborhood flows by area-overlap weighting.June was selected as a neutral month and had the best goodness-of-fit among the months considered after excluding special months.Source: Aggregated mobile phone data flows · p. 6
- 61% of origin–destination pairs were initially null on average; filtering out pairs with zeros on both weekdays and weekends reduced the share to 53%.Counts below 50 people in a time frame were reported as zero for anonymization.Source: Aggregated mobile phone data flows · pp. 6–7
- The resulting flows were concentrated between contiguous neighborhoods and suggested a northern cluster and a downtown–southern cluster extending around the port and airport highway.Source: Aggregated mobile phone data flows · p. 7, Figure 2 · p. 7
Accessibility in 15 minutes
- Accessibility is defined for each neighborhood and amenity category through the number of amenities reachable by residents within a 15-minute walk.The census provides approximate inhabitant addresses for measuring reachable amenities.Source: Accessibility in 15 minutes · p. 8
- The accessibility measure uses 2018 OpenStreetMap points of interest and the city’s walkable street network for amenity categories including education and entertainment.Source: Accessibility in 15 minutes · p. 8
- Central neighborhoods concentrate reachable amenities, while amenity diversity follows the opposite pattern and excludes public transport from its Shannon measure.Accessibility is assessed by amenity category and summarized alongside diversity across neighborhoods.Source: Accessibility in 15 minutes · p. 9, Figure 3 · p. 10
Global mobility patterns from a gravity model
- The global flow model uses a negative-binomial gravity framework, extending population and distance predictors with neighborhood attributes and origin–destination differences in accessibility and diversity.The negative-binomial distribution is used for count data to control over-dispersion.Source: Global mobility patterns from a… · pp. 11–12
Regression (GWR)
- Geographically weighted regression fits a kernel-weighted regression for each spatial unit, with flow origins chosen to examine what motivates residents to move or stay.Weights depend on each observation’s distance from the unit and a bandwidth; the authors focus on origin neighborhoods rather than destination patterns.Source: Regression (GWR) · p. 13
- GWR produces regression-factor vectors for each origin-destination pair, which can be aggregated by neighborhood to characterize local mobility and accessibility.Pairs sharing an origin receive the same regression results because they use the same spatial kernel.Source: Regression (GWR) · p. 13
- Spatial variation in GWR coefficient vectors would support the “city of cities” assumption, with neighborhoods exhibiting distinct mobility patterns.Little spatial variation would instead refute that assumption.Source: Regression (GWR) · p. 13
Relationship between local mobility and rental prices
- Spearman correlations between neighborhood rental attributes and local mobility variables assess the direction and strength of their relationship.This extends a static view based on neighborhood contiguity and land-use designations by considering how people connect neighborhoods through mobility.Source: Relationship between local mobility… · p. 14
Results
The results examine how mobility relates to neighborhood characteristics across Barcelona, considering both global and geographically varying local patterns. They also describe correlations between neighborhood attributes, local mobility factors, and rental prices across the city’s diverse functional areas.
Results
- June was selected for the remaining analysis because August and December had the lowest AICc values but coincide with holiday-related population declines.The AICc ranking was August, December, then June.Source: Results · p. 15
- In the global model, log(distance) has β = −2.11 and a negative effect, so longer distances between neighborhoods imply smaller flows; to other neighborhood has β = 1.56 and a positive effect, implying flows tend to go to other neighborhoods.Δeducation, Δretail, and Δfinance are positive.Source: Results · p. 15
- The local model has a lower AICc than the global model and reveals wide geographic variation in how neighborhood factors relate to mobility flows.Local estimates range from negative to positive for many factors; only five variables are consistently signed, including distance and visiting another neighborhood.Source: Results · p. 16, Table 1 · p. 17
- Education and retail accessibility differences have geographically varying associations with mobility, including opposite patterns across central, coastal, western, northern, and north-western neighborhoods.Education disparities are associated with mobility out of some central and northern neighborhoods but reduced mobility in others; retail disparities promote mobility from central and eastern coastal areas but are negatively associated with it in western and northern areas.Source: Results · p. 18
- Large differences in age profiles are associated with mobility from inland eastern neighborhoods, where areas with older populations tend to experience higher mobility.Source: Results · p. 18
- Retail, education, and food amenities shape local mobility patterns in most neighborhoods, with retail accessibility differences among the most consistently mobility-increasing factors.Food accessibility differences are associated with reduced mobility, which the authors suggest may reflect food amenities concentrated in the city center; entertainment and religion accessibility differences in north-western neighborhoods are associated with increased mobility.Source: Results · p. 18
- Significant neighborhood-attribute correlations with rent were positive, from access to health venues (ρ = 0.44) to food venues (ρ = 0.70).Access to religion, public transport, amenity diversity, and demographic variables related to population size, age, gender, and migration were exceptions.Source: Results · p. 20, Figure 7 · p. 21
- Spearman correlations with rental price per square meter were positive for ∆% women (ρ = 0.51), ∆food (ρ = 0.48), log(origin population) (ρ = 0.47), log(destination touristic attractiveness) (ρ = 0.46), and ∆retail (ρ = 0.44).Three variables relate to tourism: access to food and retail venues and reaching touristic places; neighborhoods where inhabitants shop and dine are more expensive.Source: Results · pp. 20–21, Figure 7 · p. 21
- Negative rent correlations were with log(destination population) (ρ = −0.57), ∆finance (ρ = −0.46), ∆health (ρ = −0.41), and ∆entertainment (ρ = −0.37).Residents tending to travel to densely populated areas had lower rents; areas where inhabitants move elsewhere for finance or health amenities were less expensive.Source: Results · p. 22
- The ∆% women factor’s positive rental-price correlation contrasts across Barcelona’s core and periphery, unlike the neighborhood share of women.The authors interpret this pattern in relation to women’s access to safe places and potential gender gaps in urban mobility.Source: Results · pp. 21–22
- The results characterize Barcelona as polycentric rather than monocentric, with distinct groups of factors influencing local mobility across diverse functional areas.The actual shapes of these multiple urban areas were not defined; the authors suggest local representations could inform planners about city functioning and ongoing phenomena.Source: Results · p. 22
- Some variables not significant for origin-destination flow inference still show variability and can inform other issues.The authors present this local variability as informative beyond flow inference.Source: Results · p. 22
Discussion and Conclusions
In Barcelona, local amenity-access effects on mobility vary substantially across neighborhoods despite global patterns, supporting spatially targeted interventions toward a 15-minute or polycentric city.
Discussion and Conclusions
- Neighborhood mobility patterns vary widely with local amenity accessibility, and differences in one or two amenity types explain patterns in most neighborhoods; food access most consistently links to reduced mobility.Source: Discussion and Conclusions · pp. 22–23
- Education and retail accessibility are positively associated with mobility globally, but locally education matters most in central and northern areas while retail stands out in coastal central neighborhoods.Both have relatively little effect in other neighborhoods.Source: Discussion and Conclusions · p. 23
- In specific neighborhoods, access to professional, health, and public transport amenities is negatively associated with mobility, potentially deterring travel to neighborhoods with poorer access to those services.Source: Discussion and Conclusions · p. 23
- The findings can guide neighborhood-specific interventions by identifying amenities and places linked to high or low mobility or resident retention, supporting 15-minute or polycentric city goals.Source: Discussion and Conclusions · p. 23
- Better access to tourist attractions may raise rental prices as readily as having an attraction; the authors suggest tourists may lower prices while proximity remains valuable.Source: Discussion and Conclusions · p. 24
- The analysis may be biased because the mobile operator’s market share favors some demographic groups, and aggregated anonymized data prevent measuring or correcting those biases.Source: Discussion and Conclusions · p. 24
- Future work could incorporate commuters and tourists, additional population and amenity-quality variables, user-generated reviews or check-ins, and explicit employment and transport-mode measures.The study models Barcelona residents and notes that visitors lack origins that map directly into the model.Source: Discussion and Conclusions · p. 24
- Further directions include time-aware models to assess policy effects and multi-city comparisons using fine-grained data to distinguish case-specific results from systematic urban patterns.The authors point to temporal extensions of GWR.Source: Discussion and Conclusions · pp. 24–25