CODEWITHSUNDEEP
stringrcharactermatch
user1665355
user1665355

Reputation: 3393

How to match by character vector with slightly different strings

I am trying to add several columns to a data.frame from an other data.frame:

The data.frame from which I want to add columns:

 head(fix)[1:2,]
Year                                         Name Moders.hjälp. Utg.Sjukvård. Antal.Fall.Moderskapshjälp. Antal.Dagar.Moderskapshjälp. Antal.Dödsfall.
1 1921                                      Allians          2003            NA                          42                         1603              43
2 1921 Bageri- och konditoriindustriarb. I Stocholm            NA            NA                          NA                           NA              10

In other words, I want to add fix[,3:ncol(fix)] to the:

head(data)[1:4,]
    Year                                                 Name Delägare.män. Delägare.kvinnor. Sjukdomsfall.män.
92  1921                                         Sbk. Allians          2416              1610               526
198 1921 Bageri- och Konditoriindustriarb. I Stockholm sbh-k.           143                13                19

by matching the Year column and Name column.

The problemis that:

The column Name in fix and data above have slightly different names (i.e. Allians VS Sbk. Allians). I can't find a correct solution that matches parts of strings to find similarities. I tried to use match but didn't succeed...

Here is dput

dput(head(fix)[1:2,])

structure(list(Year = c(1921L, 1921L), Name = c("Allians", "Bageri- och konditoriindustriarb. I Stocholm"
), Moders.hjälp. = c(2003, NA), Utg.Sjukvård. = c(NA_integer_, 
NA_integer_), Antal.Fall.Moderskapshjälp. = c(42L, NA), Antal.Dagar.Moderskapshjälp. = c(1603L, 
NA), Antal.Dödsfall. = c(43L, 10L)), .Names = c("Year", "Name", 
"Moders.hjälp.", "Utg.Sjukvård.", "Antal.Fall.Moderskapshjälp.", 
"Antal.Dagar.Moderskapshjälp.", "Antal.Dödsfall."), row.names = 1:2, class = "data.frame")

dput(head(data)[,c(1:2,11:13)])

structure(list(Year = c(1921L, 1924L, 1921L, 1924L, 1921L, 1924L
), Name = c("Sbk. Allians", "Sbk. Allians", "Bageri- och Konditoriindustriarb. I Stockholm sbh-k.", 
"Bageri- och Konditoriindustriarb. I Stockholm sbh-k.", "Bergsunds verkstads arbetares sbk", 
"Bergsunds verkstads arbetares sbk"), Delägare.män. = c(2416L, 
3896L, 143L, 129L, 280L, 289L), Delägare.kvinnor. = c(1610L, 
4300L, 13L, 13L, 2L, NA), Sjukdomsfall.män. = c(526L, 1084L, 
19L, 34L, 100L, 97L)), .Names = c("Year", "Name", "Delägare.män.", 
"Delägare.kvinnor.", "Sjukdomsfall.män."), class = "data.frame", row.names = c(92L, 
93L, 198L, 199L, 222L, 223L))

Greatful for any proposals!

Upvotes: 1

Views: 435

Answers (1)

EDi
EDi

Reputation: 13280

You can use agrep:

sapply(data$Name, function(x) agrep(x, fix$Name, max.distance=0.4))

which matches data$Name with fix$Name. You could also play around with max.distance (perhabs in a loop). Afterwards you can merge/index/etc what you want with the matches...

Update

Something along these lines should do the job for you:

# match
matches <- sapply(data$Name, function(x) agrep(x, fix$Name, max.distance=0.4))
# clean match
matches_cleaned <- sapply(matches, function(x) ifelse(length(x) > 0,x, NA))
# add matched names to data
data$fix_names <- fix$Name[matches_cleaned]

# merge
merge(data, fix, by.x = c('Year', 'fix_names'), by.y = c('Year', 'Name'))

Upvotes: 4

Related Questions