catchmentACS: Isochrone-Based Area-Weighted ACS Aggregation

catchmentACS computes estimates for drive-time areas (isochrones) around sites, such as Pre-K classrooms, by combining the American Community Survey (ACS) 5-year estimates of the census tracts that overlap each area. The tract estimates are weighted by area, and cacs_intersect_weight() describes the weights and what they assume. The five rates in cacs_acs_default_rates are ratios of the weighted counts. Each estimate has a margin of error, the half-width of its confidence interval (90 percent by default). cacs_propagate_moe() and cacs_derive_rates() describe how the margins of error are computed and what they assume. Every result carries a record of the settings and inputs behind it, which cacs_describe() prints.

Details

cacs_run() runs five steps in one call, and each step can also be called on its own:

1. cacs_acs_prefetch() downloads the ACS data for one state with tidycensus, which needs a Census API key.

2. cacs_isochrone() builds the drive-time areas with the Open Source Routing Machine (OSRM, the default) or openrouteservice, which needs an API key.

3. cacs_intersect_weight() combines the tract estimates for each area.

4. cacs_propagate_moe() computes the margins of error at a chosen confidence level.

5. cacs_derive_rates() computes the rates and their margins of error.

cacs_run() skips the download when ACS data are supplied through acs, and the routing when drive-time areas are supplied through precomputed_isochrones. The provider values "mapbox" and "r5r" and the weight_method value "population" are accepted but not implemented yet. Building drive-time areas with Mapbox or r5r, or weighting by population, gives an error.

The articles (vignettes) below are on the package website, https://joonho112.github.io/catchmentACS/articles/, and can also be opened with vignette() when the package was installed with its vignettes.

• getting-started: a first run on the bundled example data.

• alabama-tutorial: the five steps one at a time, then tables for reports.

• providers: routing services, API keys, and runs without an internet connection.

• visual-walkthrough: maps of each step for one site.

• methodology: how the estimates and margins of error are computed.

• theory-spatial-aggregation, theory-moe-propagation, and theory-derived-rates: area weighting, margins of error, and rates in detail.

• porting-v01-to-v03, porting-v03-to-v04, and porting-v04-to-v05: what to change in code written for an earlier version.

See also

• summary.cacs_run_result() and cacs_describe(): summaries of a result and of how it was produced.

• cacs_plot_site_pipeline(): maps of one site's drive-time areas, tracts, estimates, and rates.

• cacs_set_cache() and cacs_cache_dir(): turning on or off the cache, where the first three steps save their results, and where it is kept.

• cacs_acs_validate() and cacs_validate_iso(): checks of ACS data and drive-time areas from other sources.

• catchmentACS-conditions: the classes of the package's messages, warnings, and errors.

Author

Maintainer: JoonHo Lee jlee296@ua.edu (ORCID) [copyright holder]

Authors:

• JoonHo Lee jlee296@ua.edu (ORCID) [copyright holder]