LoginSign Up
← Back to all sparks
U

usmapdata

INFRA · APIS
Velocity2.5

Mapping Data for 'usmap' Package

github.com/pdil/usmapdata ↗

usmapdata ships its 2025 shapefiles on the year-indexed model it adopted in 0.4.0.

r-packagesgeospatialcensus-datacartographydata-packages
◆Current state
usmapdata supplies the boundary data behind usmap's plotting functions. Since 0.4.0 it has been year-indexed: us_map(data_year = ) selects a vintage, and each Census release is added as its own year with older ones still reachable. 1.1.0 adds 2025. 1.0.0 closed the package's longest-standing gap by adding Puerto Rico, retroactively across every vintage.
◆Where it's heading
The package has settled into a predictable rhythm — one shapefile vintage per year, with structural change rare and clustered. The two changes that mattered were data_year in 0.4.0, which turned a single-vintage dataset into a time series, and the tibble-to-data-frame switch in 0.6.0 that reduced what downstream callers have to depend on.
◆Prediction
The stated policy — each year added going forward, previous years reachable through data_year — points to a 2026 vintage as the next release. Nothing in these notes suggests further change to the data model.

◆Recent moves

  1. 23d ago

    2025 Census shapefiles added

    The annual vintage drop, exactly the cadence the data_year model was built to support. One line of notes, but it is the package's core deliverable and the reason downstream maps can match the year of the data plotted on them.

    View source ↗
  2. 0y ago

    Puerto Rico added across every map vintage

    Puerto Rico finally lands in us_map() and fips_data(), and is backfilled into every data_year rather than only the newest, so the coverage change is not gated on upgrading your vintage. include also now takes precedence over exclude when a region appears in both.

    View source ↗
  3. 1y ago

    us_map() returns a data frame instead of a tibble

    A deliberate narrowing of the return type: callers who want a tibble now convert explicitly. It is a breaking change for code that assumed tibble printing or subsetting, and it trims what the data package makes downstream users carry.

    View source ↗
  4. 1y ago

    2024 Census shapefiles added

    The 2024 vintage, arriving on the same annual cadence as 1.1.0's 2025 drop. Routine by design — the data_year machinery is what makes adding a year a one-line release.

    View source ↗
  5. 1y ago

    data_year turns the package into a multi-vintage archive

    ⚡ SPARK

    The release that set the shape of everything after it: us_map() gained data_year, converting a single-current-map dataset into an addressable archive of vintages. Every later shapefile release, including 2024 and 2025, is this decision being executed.

    View source ↗
  6. 2y ago

    Map data moves to 2023 shapefiles

    A wholesale swap to the 2023 shapefiles, made before data_year existed — so at this point updating the vintage meant replacing the only map available. The contrast with the later additive releases is the clearest measure of what 0.4.0 changed.

    View source ↗