Replace Dataframe Column Values with User Defined Function in R - r

Question

I have a grouped set of values in a column I am trying to replace with a since value
col1
a
a;a;b;c
c;b;a
NA
b;b;b
I want to replace all values with either mixed or the single present value if for example a;a;a;a becomes a
Expected Output
col1
a
Mixed
Mixed
NA
b
Code
grouping = function(x){
y = as.list(strsplit(x, ";")[[1]])
#select first element, and test if each is the same element.
z = ""
for (i in 1:length(y)){
if (as.character(y[1]) != as.character(y[i])) {
z = 'mixed'
break
} else {
z = as.character(y[1])
}
}
return(z)
}
db %>%
select(col1) %>%
mutate(
test = grouping(col1)
)
I have tried it a few different ways and either end up with it not working at all or giving the value a for everything

Answer 1

A base R option via defining a user function f
f <- function(x) ifelse(length(u <- unique(unlist((strsplit(x, ";"))))) > 1, "Mixed", u)
such that
> transform(df, col1 = Vectorize(f)(col1))
col1
1 a
2 Mixed
3 Mixed
4 <NA>
5 b

Answer 2

You can also consider this for your function and use base R:
#Function
myfun <- function(x)
{
y <- unlist(strsplit(x, ";"))
if(length(unique(y))==1)
{
z <- unique(y)
} else
{
z <- 'Mixed'
}
}
#Apply
df$New <- apply(df,1,myfun)
Output:
df
col1 New
1 a a
2 a;a;b;c Mixed
3 c;b;a Mixed
4 <NA> <NA>
5 b;b;b b
Some data used:
#Data
df <- structure(list(col1 = c("a", "a;a;b;c", "c;b;a", NA, "b;b;b")), class = "data.frame", row.names = c(NA,
-5L))

Answer 3

We can extract the substring from the 'col1' which are letters, check the number of distinct elements with n_distinct, use case_when to change those which have more one unique elements to 'Mixed'
library(dplyr)
library(stringr)
library(purrr)
df1 %>%
mutate(col1 = case_when(map_dbl(str_extract_all(col1,
"[a-z]"), n_distinct) >1 ~ "Mixed",
is.na(col) ~ NA_character_,
TRUE ~ substr(col1, 1, 1)))
-output
# col1
#1 a
#2 Mixed
#3 Mixed
#4 <NA>
#5 b
Or another option is to split the column by the delimiter with separate_rows, and do a group by row_number to summarise elements having more than one row (after the distinct) to be 'Mixed'
library(tidyr)
df1 %>%
mutate(rn = row_number()) %>%
separate_rows(col1) %>%
distinct() %>%
group_by(rn) %>%
summarise(col1 = case_when(n() > 1 ~ 'Mixed', TRUE ~ first(col1)),
.groups = 'drop') %>%
select(-rn)
-output
# A tibble: 5 x 1
# col1
# <chr>
#1 a
#2 Mixed
#3 Mixed
#4 <NA>
#5 b
Or using base R with a compact option
v1 <- gsub("([a-z])\\1+", "\\1", gsub(";", "", df1$col1))
replace(v1, nchar(v1) > 1, "Mixed")
#[1] "a" "Mixed" "Mixed" NA "b"
The issue in the OP's function is that it is extracting only the first [[1]] list element
as.list(strsplit(x, ";")[[1]])
as strsplit returns a list with length equal to the number of rows of the initial data. So, basically by selecting only the first, it is recycled
data
df1 <- structure(list(col1 = c("a", "a;a;b;c", "c;b;a", NA, "b;b;b")),
class = "data.frame", row.names = c(NA,
-5L))

Answer 4

You can write the grouping function as :
grouping <- function(x) {
sapply(strsplit(x, ';'), function(x)
if(length(unique(x)) == 1) unique(x) else 'Mixed')
}
db$test <- grouping(db$col1)
db
# col1 test
#1 a a
#2 a;a;b;c Mixed
#3 c;b;a Mixed
#4 <NA> <NA>
#5 b;b;b b