Match a loanbook to asset-based company data (abcd) by the name_* columns

match_name() scores the match between names in a loanbook dataset (columns can be name_direct_loantaker, name_intermediate_parent* and name_ultimate_parent) with names in an asset-based company data (column name_company). The raw names are first internally transformed, and aliases are assigned. The similarity between aliases in each of the loanbook and abcd is scored using stringdist::stringsim().

Usage

match_name(
  loanbook,
  abcd,
  by_sector = TRUE,
  min_score = 0.8,
  method = "jw",
  p = 0.1,
  overwrite = NULL,
  join_id = NULL,
  ...
)

Arguments

loanbook, abcd: data frames structured like r2dii.data::loanbook_demo and r2dii.data::abcd_demo.
by_sector: Should names only be compared if companies belong to the same sector?
min_score: A number between 0-1, to set the minimum score threshold. A score of 1 is a perfect match.
method: Method for distance calculation. One of c("osa", "lv", "dl", "hamming", "lcs", "qgram", "cosine", "jaccard", "jw", "soundex"). See stringdist::stringdist-metrics.
p: Prefix factor for Jaro-Winkler distance. The valid range for p is 0 <= p <= 0.25. If p=0 (default), the Jaro-distance is returned. Applies only to method='jw'.
overwrite: A data frame used to overwrite the sector and/or name columns of a particular direct loantaker or ultimate parent. To overwrite only sector, the value in the name column should be NA and vice-versa. This file can be used to manually match loanbook companies to abcd.
join_id: A join specification passed to dplyr::inner_join(). If a character string, it assumes identical join columns between loanbook and abcd. If a named character vector, it uses the name as the join column of loanbook and the value as the join column of abcd.
...: Arguments passed on to stringdist::stringsim().

Value

A data frame with the same groups (if any) and columns as loanbook, and the additional columns:

id_2dii - an id used internally by match_name() to distinguish companies
level - the level of granularity that the loan was matched at (e.g direct_loantaker or ultimate_parent)
sector - the sector of the loanbook company
sector_abcd - the sector of the abcd company
name - the name of the loanbook company
name_abcd - the name of the abcd company
score - the score of the match (manually set this to 1 prior to calling prioritize() to validate the match)
source - determines the source of the match. (equal to loanbook unless the match is from overwrite

The returned rows depend on the argument min_value and the result of the column score for each loan: * If any row has score equal to 1, match_name() returns all rows where score equals 1, dropping all other rows. * If no row has score equal to 1,match_name() returns all rows where score is equal to or greater than min_score. * If there is no match the output is a 0-row tibble with the expected column names -- for type stability.

Package options

r2dii.match.sector_classifications: Allows you to use your own sector_classififications instead of the default. This feature is experimental and may be dropped and/or become a new argument to match_name().

Assigning aliases

The transformation process used to compare names between loanbook and abcd datasets applies best practices commonly used in name matching algorithms:

Remove special characters.
Replace language specific characters.
Abbreviate certain names to reduce their importance in the matching.
Spell out numbers to increase their importance.

Handling grouped data

This function ignores but preserves existing groups.

Examples

if (FALSE) {
library(r2dii.data)
library(tibble)

# Small data for examples
loanbook <- head(loanbook_demo, 50)
abcd <- head(abcd_demo, 50)

match_name(loanbook, abcd)

match_name(loanbook, abcd, min_score = 0.9)

# Use your own `sector_classifications`
your_classifications <- tibble(
  sector = "power",
  borderline = FALSE,
  code = "D35.11",
  code_system = "XYZ"
)

# match on LEI
loanbook <- tibble(
  sector_classification_system = "XYZ",
  sector_classification_direct_loantaker = "D35.11",
  id_ultimate_parent = "UP15",
  name_ultimate_parent = "Won't fuzzy match",
  id_direct_loantaker = "C294",
  name_direct_loantaker = "Won't fuzzy match",
  lei_direct_loantaker = "LEI123"
)

abcd <- tibble(
  name_company = "alpine knits india pvt. limited",
  sector = "power",
  lei = "LEI123"
)

match_name(loanbook, abcd, join_by = c(lei_direct_loantaker = "lei"))

restore <- options(r2dii.match.sector_classifications = your_classifications)

loanbook <- tibble(
  sector_classification_system = "XYZ",
  sector_classification_direct_loantaker = "D35.11",
  id_ultimate_parent = "UP15",
  name_ultimate_parent = "Alpine Knits India Pvt. Limited",
  id_direct_loantaker = "C294",
  name_direct_loantaker = "Yuamen Xinneng Thermal Power Co Ltd"
)

abcd <- tibble(
  name_company = "alpine knits india pvt. limited",
  sector = "power"
)

match_name(loanbook, abcd)

# Cleanup
options(restore)
}

Match a loanbook to asset-based company data (abcd) by the `name_*` columns