Skip to contents

Infer column types using readr, then layer factor detection

Source: R/auto_parse_types.R
auto_parse_types.Rd

use readr::type_convert() to infer non-factor types (numeric, integer, date, datetime, logical). then propose factors via low-cardinality rules. protect id-like names and leading-zero codes. return the dataset by default, and include the metadata plan only when requested.

Usage

auto_parse_types(
  data,
  max_levels = 50,
  max_unique_ratio = 0.2,
  protect_patterns = c("id$", "uid$", "code$", "ref$", "key$"),
  keep_leading_zero_chars = TRUE,
  prefer_logical_for_binary = TRUE,
  apply = TRUE,
  return = c("data", "both", "plan")
)

Arguments

data

data.frame or tibble.

max_levels

integer. max distinct values for factor. default 50.

max_unique_ratio

numeric (0, 1]. max unique/n for factor. default 0.2.

protect_patterns

character regexes for protected names. default c("id$", "uid$", "code$", "ref$", "key$").

keep_leading_zero_chars

logical. keep character if any value has leading zeros in digit-only strings. default TRUE.

prefer_logical_for_binary

logical. kept for api compatibility, not used when delegating to readr. default TRUE.

apply

logical. if TRUE, apply factor conversions on top of parsed types. default TRUE.

return

one of "data", "both", "plan". default "data".

Value

invisible object depending on return:

  • "data": tibble of parsed data (and factors if apply = TRUE)

  • "both": list(plan = tibble, data = tibble as above)

  • "plan": tibble only

Examples

df <- tibble::tibble(
  id  = c("001", "002", "003"),
  sex = c("M", "F", "F"),
  age = c("1", "2", "3"),
  dt  = c(
    "2024-01-01 12:00:00",
    "2024-01-02 00:00:00",
    "2024-01-03 01:02:03"
  )
)

# parsed types + inferred factors
dat <- auto_parse_types(df, apply = TRUE, return = "data")

# dataset and plan
both <- auto_parse_types(df, apply = TRUE, return = "both")
both$plan |> dplyr::select(name, current_type, proposed_type, rule)
#> # A tibble: 4 × 4
#>   name  current_type proposed_type rule             
#>   <chr> <chr>        <chr>         <chr>            
#> 1 id    character    character     protected_by_name
#> 2 sex   character    character     readr:character  
#> 3 age   character    integer       readr:integer    
#> 4 dt    character    POSIXct       readr:POSIXct    
dplyr::glimpse(both$data)
#> Rows: 3
#> Columns: 4
#> $ id  <chr> "001", "002", "003"
#> $ sex <chr> "M", "F", "F"
#> $ age <int> 1, 2, 3
#> $ dt  <dttm> 2024-01-01 12:00:00, 2024-01-02 00:00:00, 2024-01-03 01:02:03