urban planner | educator | maker
js6676@columbia.edu
cv  


projects
urban planningurban technologytransportation
climate
urban design
master’s thesis



blog 
about 





















© Jui Shih, 2023 - 2026




Urban Informatics:
Recreational Access X Social Vulnerability in NYC




Subject(s)
Public health, Public Space, Health Equity, Urban Planning, 
Spatial Analysis

Course
Urban Informatics & Analytics

Tools
Python

Project Location    
New York, NY

Completed Period
December 2025



How evenly is access to parks and recreational infrastructure distributed in New York City, you might ask? This is a two-part spatial analytics project investigating how equitably outdoor recreation is located in NYC. Using OpenStreetMap, American Community Survey (ACS), and Python, I mapped parks, playgrounds, bike infrastructure, and pedestrian networks to evaluable walkable access to recreational amenities. I then cross-checked it with socio-economic indicators such as poverty, youth population, disability percentages across the 5 boroughs to examine whether neighborhoods with higher social vulnerability experience lower access to recreation facilities. 

The project integrates GIS techniques such as spatial joins, buffering, centroid analysis, and interactive web mapping while critically evaluating data quality, methodological assumptions, and opportunities for improving accessibility metrics. In particular, I used Python libraries such as GeoPandas and Folium to transform spatial datasets into interactive visualizations and perform geospatial analyses. 


Data Design


Part I: Identify where outdoo recreation infrastructure exists and how it connects to pedestrian networks. 

I selected the following OSM tags as they represent a comprehensive range of places used for recreational purposes.  This study includes bike paths in the mix as outdoor recreation infrastructure. In cities as dense and urbanized as NYC, cycling infrastructure should be within walkable access for residents, and ideally pedestrian paths and bike paths should be interlinked to form an “active mobility network” Additionally, including the tags for pedestrian accessibility selected for this study.

Key
leisure
leisure
leisure
leisure
leisure
highway
cycleway 
cycleway
cycleway
bicycle
highway
highway
highway
footway
Value
park
pitch
playground
track
common
cycleway
lane
track
shared_land
designated
footway
pedestrian
living_street
sidewalk
Description
A park in an urban/municipal setting created for recreation & relaxation.
An area designed for practicing particular sport, has designated markings.
An area designated for children to play.
Tracks for running, cycling, and other non-motorized racing.
Land over which the public has general rights of use for certain leisure activities.
Routes that are designated for bicycles.
A lane is a route that lies within the roadway.
A track provides a route that is separated from traffic.
These are lanes where cyclists share with motor vehicles.
As the name suggests, these are designated for cycling.
Designated footpaths mainly/ exclusively for pedestrians.
Roads use mainly or exclusively for pedestrians.
Residential streets where pedestrians have legal priority over cars.
Sidewalks that run along residential roads. 


Part II: Identify whether socioeconomic risk is associated with reduced access to recreational infrastructure in NYC and where those disparities are most visible. 

I drew 3 variabels from U.S. Census Amercian Community Survey (ACS) data that I believe to be representative of socioeconomic vulnerabilities and potential barriers to outdoor recreation access 

These indicators represent populatios who may rely more on nearby recrational spaces yet may face greater constrains in reaching them. Youths require proximate outdoor environments for healthy physical and cognitive development, while households in poverty often experience limited mobility options as well as people with disabilities face physical barriers that may restrict access to parks and paths.
   ACS Datasets Sourced

  • DP05_0019PE: % People under 18
  • S1701_C03_001E: % People living under poverty level
  • S1810_C03_001E: % People with disabilities






Maps & Visualization


Part I: 



Part II: Recreation Accessibility X Social Vulnerability in NYC

I built the below interactive map in Python using Folium library that renders Leaflet.js web map. It overlays NYC parks and recreation infrastructures/ sites on a choropleth of the Social Vulnerability Index (0-100). Parks, playgrounds, and recreation facilities/ infrastructure were plotted as markers with hovering ability to show  each each site’s name and accessibility score. Feel free to explore!






Method

Findings


For the first part, I assessed the relationship between recreation and accessibility in Python by using pandas, geopandas, shapely, matplotlib and contextily. Each recreation feature chosen via OSM was buffered by 400 meters, corresponding to a 5 minute walk. Then they were spatially joined with pedestrian networks, and the features whose buffers intersected with at least one pedestrian segment were deemed as “pedestrian accessible”.

I then calculated the percentage of accessibility by borough using the subset figure and the total number of recreation features. Visually, I also overlaid the pedestrian network with recreation features on a basemap to see how they intersect. 

To compare accessibility with socioeconomic risk, the recreation geometries were converted to centroid to create a simplified point that can be assigned to a single tract, otherwise some (larger) polygons may intersect with multiple census tract. The centroid points were then spatially joined to census tracts, linking each recreation feature to a corresponding tracts GEOID, which was then flagged for accessibility with 1 being within pedestrian reach and 0 without. As this was census-tract tagged, I was then able to compare it with ACS socioeconomic variables I had selected (under 18, individuals below poverty and individuals with disabilities), each were then standardized using z-scores and averaged to create equal-weighted, and finally normalized to 0-100, with higher values meaning greater socioeconomic risk.



Implications & Limitations


My findings indicate that spatial injustice in NYC is less about the overall supply of recreation as there are many, and more about uneven, localized gaps. Relying on the percentage of recreation feature alone quite obscures the variation in sizes and quality of outdoor recreation. My analysis was further constrained by using a simple buffer measure and (possibility) the ACS datasets used. Further tweaking of the analysis such as area-based metrics might provide a more accurate picture of spatial equity.



In general when viewing the folium map with the overlay of recreation infrastructure, high socioeconomic risk and low recreational access tend to overlap in tracts located in the Bronx, some areas in eastern Queens and portions of central and southern Brooklyn. These areas have higher numbers of youth, poverty, and disability, while also have fewer recreation features. In that sense, the tract level analysis reveals a more nuanced picture compared to the borough-level. However upon further exploration, I found some discrepancies. 





The top screenshot captures an area in the south of Bronx. Although the tract is technically classified as “accessible”, very few outdoor recreation features exist. The scale bar helps to show that the area is wider than 800 meter radius, which is equivalent to the distance of a 5 minute walk. 

The bottom image shows a low vulnerability tract that similarly lacks nearby recreation infrastructure. The overlay of recreation features on the tract map helps to reveal these localized gaps more clearly, at the same time led me to question my methodology used in computing the recreation accessibility.



 BACK