Aggregate or interpolate ACS data to custom geographies — interpolate

Aggregate or interpolate ACS data from source geographies to user-defined target geographies. Supports two modes:

Complete nesting (weight = NULL): Each source geography maps entirely to one target geography. Count variables are summed, percentages are recalculated from summed components, and intensive variables (medians, averages) are computed as population-weighted averages.

Fractional allocation (weight = "column_name"): Source geographies can be split across multiple targets using crosswalk weights. Count variables and MOEs are multiplied by the weight before summing. Percentages are recalculated from interpolated components. Intensive variables use the allocated population as weights.

MOE propagation uses Census Bureau approximation formulas throughout. Crosswalk weights are treated as constants (no sampling error).

Usage

interpolate_acs(
  .data,
  target_geoid,
  weight = NULL,
  crosswalk = NULL,
  source_geoid = "GEOID",
  weight_variable = "total_population_universe"
)

Arguments

.data: A dataframe returned from compile_acs_data(). Must have a codebook attribute attached.
target_geoid: Character. Column name for target geography identifiers. Must exist in .data or in crosswalk. The result renames this column to GEOID.
weight: Character or NULL. When NULL (default), assumes complete nesting where each source geography maps entirely to one target. When a column name is provided, performs fractional allocation using that column as weights. Weights should sum to approximately 1 per source geography.
crosswalk: A data frame containing the crosswalk mapping. Optional in both modes. When provided, joined to .data via source_geoid before processing. Must include columns for source_geoid and target_geoid (and weight if fractional allocation is used).
source_geoid: Character. Column name for source geography identifiers. Must exist in .data (and in crosswalk if provided). Default is "GEOID".
weight_variable: Character. Variable name used for population-weighted averages of intensive variables (medians, averages, etc.). Default is "total_population_universe".

Value

A dataframe aggregated to target geographies with recalculated estimates, MOEs, and percentages. A modified codebook is attached as an attribute.

Examples

if (FALSE) { # \dontrun{
# First, create tract-level data
tract_data = compile_acs_data(
  tables = c("race", "snap"),
  years = 2022,
  geography = "tract",
  states = "DC"
)

# Complete nesting: each tract belongs to exactly one neighborhood
tract_data$neighborhood = c("Downtown", "Downtown", "Uptown", ...)
neighborhood_data = interpolate_acs(
  .data = tract_data,
  target_geoid = "neighborhood"
)

# Fractional allocation with a crosswalk
crosswalk = data.frame(
  GEOID = c("11001000100", "11001000100", "11001000201"),
  neighborhood = c("Downtown", "Chinatown", "Downtown"),
  alloc_weight = c(0.6, 0.4, 1.0)
)

neighborhood_data = interpolate_acs(
  .data = tract_data,
  target_geoid = "neighborhood",
  weight = "alloc_weight",
  crosswalk = crosswalk
)
} # }