Skip to contents

Build the EJSCREEN web-app "thresholds" layers (P1..P100 hit counts)

Source: R/calc_ejscreen_threshold_layers.R
calc_ejscreen_threshold_layers.Rd

Produces the four "threshold map widget" layers requested for the EJSCREEN web app in issue #395: us_ejindexes, state_ejindexes, us_supplemental, and state_supplemental. Each layer has one row per block group containing the per-index percentile ranks plus 100 columns P1..P100 giving, for that block group, how many of its EJ-index percentile ranks land at each integer percentile (e.g. if a block group is at the 4th percentile for P_D2_OZONE and no other D2 index rank is 4, then P4 = 1).

Usage

calc_ejscreen_threshold_layers(
  pctiles,
  id_col = "bgfips",
  layers = c("us_ejindexes", "state_ejindexes", "us_supplemental", "state_supplemental"),
  cols_us_ej = names_ej_pctile,
  cols_state_ej = names_ej_state_pctile,
  cols_us_supp = names_ej_supp_pctile,
  cols_state_supp = names_ej_supp_state_pctile,
  out_dir = NULL
)

Arguments

pctiles

A data.frame, one row per block group, containing id_col and the EJ-index percentile-rank columns – for example the output of calc_ejscreen_export() (which holds P_D2_* and P_D5_*), or a merge of the national and state exports so all four column sets are present.

id_col

Name of the block-group id column to carry through. Default "bgfips".

layers

Which layers to build; any of "us_ejindexes", "state_ejindexes", "us_supplemental", "state_supplemental".

cols_us_ej, cols_state_ej, cols_us_supp, cols_state_supp

Character vectors of the percentile-rank column names for each layer. Defaults are the EJAM name vectors names_ej_pctile, names_ej_state_pctile, names_ej_supp_pctile, names_ej_supp_state_pctile.

out_dir

Optional directory; if supplied, each layer is written there as <layer>.csv.

Value

A named list of data.frames, one per requested layer (the id column + the percentile-rank columns + P1..P100).

Details

Percentile ranks are rounded to integers and clamped into 1..100 (a rank that rounds to 0 is counted in P1, anything above 100 in P100, and NA ranks are not counted), then the number of indexes at each integer percentile is tallied per block group in a single pass over the data. For ranks already in 1..100 this equals the colcounter() identity colcounter(x, k) - colcounter(x, k + 1) (count at-or-above k minus count at-or-above k + 1), i.e. it is the per-block-group analogue of the EXCEED_COUNT_* fields in calc_ejscreen_export().

The four layers differ only in which set of percentile columns they tabulate:

In the annual pipeline these layers are produced by the ejscreen_export stage of calc_ejscreen_dataset() when include_ejscreen_export is TRUE: it calls calc_ejscreen_threshold_layers_from_exports() on the national and state exports and saves them under the registered stage names ejscreen_threshold_us_ejindexes, ejscreen_threshold_us_supplemental, ejscreen_threshold_state_ejindexes, and ejscreen_threshold_state_supplemental. (Issue #395 part (a) — the US/State percentile CSVs — is the existing ejscreen_export / ejscreen_export_statepct; part (c) — the ACS-by-geography layers — is calc_acs_by_geography().)

See also

colcounter() calc_ejscreen_export() calc_ejscreen_dataset()