Week 4 Notes - Spatial Data & GIS

Published

September 29, 2025

Key Concepts Learned

  • Spatial Data Fundamentals
    • Vector data model and sf objects
    • ESRI Shapefile format: .shp (geometry – lat/long); .dbf (attribute); .sbx (binary spatial index file)
  • Spatial Operations
    • Spatial subsetting and filtering
    • Spatial joins and relationships
    • Measuring distances and areas
  • Geometry Operations
    • Buffers, intersections, and unions
    • Spatial aggregation
  • Coordinate Reference Systems (CRS)
    • Understanding projections & transforming between CRS
    • Geographic Coordinate Systems (GCS) utilize lat/long coordinates with angular units in decimal degrees
      • WGS84 (EPSG:4326):GPS standard, global coverage, good for web mapping, data sharing
    • Projected Coordinate Systems (PCS) utilize X/Y coordinates on a flat plane with units in metres/feet

      • Web Mercator (EPSG:3857): Web mapping standard, heavily distorts areas near poles

      • State Plane / UTM zones: local accuracy, different zones for different regions, optimized for specific geographic areas

Coding Techniques

Read Spatial Data: st_read

Spatial Subsetting: extract features based on spatial relationships

  • Which counties border Allegheny? → st_touches looks for neighbors
  • Which tracts are IN Allegheny? → st_within
  • Which tracts overlap a metro area? → st_intersects includes the reference feature itself
  • If no .predicate specified, uses st_intersects

Code 
# Key Functions
st_filter(data_to_filter, reference_geometry, .predicate = relationship)

# .predicate = ""
st_intersects()
st_touches()
st_within()

Checking and Setting CRS

  • To simply check current CRS st_crs(pa_counties)
  • To set CRS (ONLY if missing) pa_counties <- st_set_crs(pa_counties, 4326)
Code 
# st_transform
census_tracts <- census_tracts %>%
  st_transform(st_crs(pa_counties))

# Transform to Pennsylvania South State Plane (good for PA analysis)
pa_counties_projected <- pa_counties %>%   st_transform(crs = 3365)

# Transform to Albers Equal Area (good for area calculations) 
pa_counties_albers <- pa_counties %>%   st_transform(crs = 5070) `

Plot Maps

Code 
ggplot(pa_counties) +
  geom_sf() +
  theme_void()

Spatial Joins

Code 
tracts_with_counties <- census_tracts %>%
  st_join(pa_counties)

Area Calculations

  • The dot (.) is a placeholder that represents the data being passed through the pipe
  • In the code example below, the dot (.) refers to pa_counties
Code 
# Convert to Sq Km
pa_counties <- pa_counties %>%
  mutate(
    area_sqkm = as.numeric(st_area(.)) / 1000000
  )

Distance and Buffers

  • buffer units depend on coordinate reference system
Code 
# 10km buffer around all hospitals
hospitals_projected <- hospitals %>%
  st_transform(crs = 3365)
st_crs(hospitals_projected)

hospital_buffers <- hospitals_projected %>%
  st_buffer(dist = 32808.4)  # 10,000 meters = 32808.4 feet

ggplot(hospital_buffers)+
  geom_sf()+
  theme_void()

# Different buffer sizes by hospital type (this is hypothetical, we don't have that column!)
hospital_buffers <- hospitals %>%
  mutate(
    buffer_size = case_when(
      type == "Major Medical Center" ~ 15000,
      type == "Community Hospital" ~ 10000,
      type == "Clinic" ~ 5000
    )
  ) %>%
  st_buffer(dist = .$buffer_size)

# Use Unions to calculate what percentage of low-income tracts have access
access_summary <- low_income_tracts %>%
  mutate(
    has_access = st_intersects(., st_union(hospital_buffers), sparse = FALSE)
  ) %>%
  st_drop_geometry() %>%
  summarize(
    total_tracts = n(),
    tracts_with_access = sum(has_access),
    pct_with_access = (tracts_with_access / total_tracts) * 100
  )

print(access_summary)

Questions & Challenges

  • No questions for this week

Connections to Policy

  • Spatial Policy Analysis Workflow

    • Load data → Get spatial boundaries and attribute data

    • Check projections → Transform to appropriate CRS

    • Join datasets → Combine spatial and non-spatial data

    • Spatial operations → Buffers, intersections, distance calculations

    • Aggregation → Summarize across spatial units

    • Visualization → Maps and charts

    • Interpretation → Policy recommendations

Reflection

  • I was surprised at how intuitive and quick it was to develop maps and visualizations using packages in R as compared to ArcGIS. I discussed this with Prof. Delmelle during Office Hours, and she shared that she has relied exclusively on R for the past two years after switching to a Macbook.

  • While ArcGIS is still a useful tool and has many cool additional features such as StoryMaps or animations, I believe that R is the way forward for core data wrangling, visualization, and machine learning work.