Skip to contents

Generate tidyverse Code for Automatic Discrepancy Cleaning and Crosswalks

Source: R/codegen.R
sv_auto_clean_rules.Rd

Inspects discrepancy detection output or pairwise column discrepancies (exact matches, fuzzy near-matches, and missing levels), automatically synthesizing clean recoding rules using dplyr::case_when(). Handles casing differences, leading/trailing whitespace discrepancies, and fuzzy string distance variations (Levenshtein distance <= 2).

Usage

sv_auto_clean_rules(
  discrepancy_result,
  df_name = "df",
  col = "status",
  default = col,
  pipe = FALSE,
  lazy = FALSE,
  df_is_code = FALSE
)

sv_code_auto_clean(
  discrepancy_result,
  df_name = "df",
  col = "status",
  default = col,
  pipe = FALSE,
  lazy = FALSE,
  df_is_code = FALSE
)

Arguments

discrepancy_result

A discrepancy result list (such as returned by sv_detect_discrepancies()) or an object containing fuzzy_matches, only_in_x, only_in_y, and exact_matches.

df_name

Name of the data frame to mutate. Defaults to "df". Can be NULL for piped or standalone expressions.

col

Column name being recoded. Defaults to "status". If NULL, attempts to extract from discrepancy_result$from_col or defaults to "status".

default

Fallback value for unmatched records. If equal to col, preserves unmatched records via .default = <col>. Set to NULL to omit the default clause.

pipe

Logical; if TRUE, emits piped syntax df %>% dplyr::mutate(...).

lazy

Logical; TRUE when df_name is an arrow or duckdb table. The code then ends in dplyr::collect() (the one collect of a lazy pipeline), the recode closes with TRUE ~ <col> (arrow cannot take .default = <col>), and df_name may be an expression such as dplyr::tbl(con, "visits") when df_is_code is TRUE. The matching of values is done in R on the values compared, so the generated code uses no trimws() or tolower().

df_is_code

Logical; TRUE when df_name is R code (such as dplyr::tbl(con, "visits")) to use as written. FALSE (the default) treats it as a table name and backtick-quotes it when it is not syntactic, as in visits (2020). Only used when lazy is TRUE.

Value

A character string of tidyverse code, or NULL if no discrepancies require cleaning.

Examples

disc <- sv_detect_discrepancies(c("male", "Female"), c("Male", "Female"))
sv_auto_clean_rules(disc, df_name = "patients", col = "gender")
#> [1] "dplyr::mutate(patients, gender = dplyr::case_when(\n    gender == \"male\" ~ \"Male\",\n    .default = gender\n  ))"