vrmatch: Voter Registration Database Snapshot Matching

In sysilviakim/voterdiffR: Evaluating the Quality of Changes in Voter Registration Databases

Description

This function performs probabilistic record linkage between all user-supplied consecutive snapshots of the voter file. Note that the default option is to exclude exact matches of all fields between two snapshots when performing the record linkage, for computational reasons. Note that for multiple matchings, the function uses a loop instead of more sophisticated measures such as purrr::map, because loading and wrangling them simultaneously will often times bring the machine crashing down.

Usage

vrmatch(date_df, exact_exclude = TRUE, sample_exact = FALSE,
  sample_id = FALSE, sample_size = NULL, sample_perc = NULL,
  block = FALSE, path_clean = "clean_df", path_changes = "changes",
  path_reports = "reports", path_matches = "matches",
  clean_prefix = "df_cleaned_", clean_suffix = "",
  exist_files = FALSE, varnames, varnames_str, varnames_num = NULL,
  varnames_id = NULL, partial.match = NULL, varnames_block = NULL,
  vars_change = NULL, n.cores = NULL, file_type = ".Rda",
  date_label = "date_label", nrow = "nrow", seed = 123, ...)

Arguments

date_df	Dataframe of list of snapshots.
exact_exclude	Whether to exclude full exact matches between snapshots when doing probabilistic record linkage. Defaults to TRUE.
sample_exact	Whether to add random samples of full exact matches to correct for underlying population's value distributions for each field. Defaults to FALSE.
sample_id	Whether to add random samples of ID matches (some changes) to correct for underlying population's value distributions for each field. Defaults to FALSE.
sample_size	Sample size of the random sample to add. Defaults to NULL.
sample_perc	Sample percentage of the random sample to add. Defaults to NULL. If both 'sample_size' and 'sample_perc' are NULL, 'sample_perc' is set to 0.01 (1 'sample_perc' is chosen over 'sample_size'.
block	Whether to employ blocking. Defaults to FALSE.
path_clean	Path to the cleaned snapshots. Defaults to "clean_df".
path_changes	Path where the extracted changes are output to. Defaults to "changes".
path_reports	Path where the summarized changes are output to. Defaults to "reports".
path_matches	Path where the match outcomes are output to. Defaults to "matches".
clean_prefix	File prefixes for cleaned snapshots. Defaults to "df_cleaned_".
clean_suffix	File suffixes for cleaned snapshots. Defaults to empty string.
exist_files	Whether previously performed match outcomes exist. Defaults to FALSE.
varnames	Variables to perform probabilistic record linkage.
varnames_str	String variables for matching.
varnames_num	Numeric variables for matching. Defaults to NULL, in which case it will be setdiff(varnames, varnames_str).
varnames_id	Voter IDs variables, if any exists, and is to be excluded from PRL when IDs match.
partial.match	Variables to be partially matched. Defaults to all varnames_str.
varnames_block	Nested list of variables or their combinations for blocking passes.
vars_change	Variables to track changes of. Defaults to NULL, which will then track all variables.
n.cores	Number of cores to parallelize the matching. Defaults to half the existing threads.
file_type	Input file types. Defulats to .Rda.
date_label	Labels for dates (i.e., snapshot IDs), in 'date_df'. Defaults to "date_label".
nrow	Name of list element which will contain the number of rows of the input list dataframes.
seed	Seed to set. Defaults to 123.
...	Other parameters for fastLink.

Value

A nested list of matched dataframes, fastLink output, and arguments.

sysilviakim/voterdiffR documentation built on June 22, 2020, 6:51 p.m.