Skip to contents

Compile Spatial Predictor Variables

compileSpatial.Rd

Detects, aggregates, and harmonizes all processed spatial datasets in geo-processed/ into a single, standardized spatial predictor file (geo_predictors.fst) and spatial dictionary (spatial).

Usage

compileSpatial()

Details

Processes geographic covariates (e.g., land use, walkability, climate) across varying temporal vintages and spatial units into uniform PUMA-level summaries.

The workflow proceeds as follows:

  • Spatial Aggregation: Processes spatial datasets in parallel across CPU cores via summarizeDataset(). Matches geographic source geometries to 2010 and 2020 PUMA boundaries using geographic concordances, aggregating metrics via weighted means (numeric/logical variables) or weighted modes (categorical variables).

  • Metadata Extraction: Extracts variable labels, data types, and observed vintage ranges into a standardized spatial dictionary dataset (spatial).

  • Dense Rank Transformation: Converts numeric predictor variables into dense integer percentile ranks (data.table::frank(..., ties.method = "dense")) within each state-PUMA-vintage grouping prior to expansion.

  • Temporal Expansion: Fills temporal gaps across years (2000 through the prior calendar year) by holding boundary vintages constant.

  • Unified Storage: Outer-joins all processed spatial datasets across PUMA-vintages and exports compressed binary datasets (geo_predictors.fst and spatial.rda).

Important: This function must be executed with your R working directory set to the root of the local fusionData project folder (e.g., setwd("path/to/fusionData")). It relies on relative directory paths (geo-processed/.