match_names: Approximate matching between character vectors

In etnite/bwardr: Brian Ward personal R package

Approximate matching between character vectors

Description

Approximate matching between character vectors

Usage

match_names(a, b, max_dist = 5, drop_incomparables = TRUE, drop_nonsubs = TRUE)

Arguments

a	A character vector
b	A character vector
max_dist	Integer - the maximum edit distance to consider for approximate string matching using the longest common substring method
drop_incomparables	Logical - whether to output names of entries in either input character vector that could not be approximately matched to any entry in the other input vector
drop_nonsubs	Logical - whether to drop entries in either input character vector that are not a full substring of any entry in the other input vector

Details

For two input character vectors, the function will first create "simplified" versions of each, in which all non-alphanumeric characters are stripped out, and all alphabetic characters are converted to uppercase. Approximate string matching is then performed using the longest common substring (LCS) method in the stringdist package. Note that this may produce one-to-multiple matchings. The "substring" column in the output dataframe will indicate whether one term is a substring fully contained within the other. For instance, "Foobar" and "Foobaz" share the substring "Fooba", but neither is a full substring of the other. In contrast, "Foobar" is a valid substring of "Foobarbaz". The difference between any two corresponding elements in the input vectors is calculated using base R's adist() function. This uses the Levenshtein distance rather than the lcs metric, though its use here is "protected" by the previous use of the LCS metric to perform the approximate string matching.

Value

A dataframe containing seven columns: The original and simplified versions of the input vectors (see details), the edit distance between the two simplified vector entries, the difference between the two simplified vector entries, and a logical column indicating whether an entry in one of the vectors is a full substring of the corresponding entry in the other

etnite/bwardr documentation built on Jan. 6, 2023, 7:12 a.m.