Separate data frame depending on one column duplicates - r

Question

I have a large data frame with a lot of rows and columns. In one column there are characters, some of them occur only once, other multiple times. I would now like to separate the whole data frame, so that I end up with two data frames, one with all the rows that have characters that repeat themselves in this one column and another one with all the rows with the charcaters that occur only once. Like for example:
One = c(1,2,3,4,5,6,7,8,9,10)
Two = c(4,5,3,6,2,7,1,8,1,9)
Three = c("a", "b", "c", "d","d","e","f","e","g","c")
df <- data.frame(One, Two, Three)
> df
One Two Three
1 1 4 a
2 2 5 b
3 3 3 c
4 4 6 d
5 5 2 d
6 6 7 e
7 7 1 f
8 8 8 e
9 9 1 g
10 10 9 c
I wish to have two data frames like
> dfSingle
One Two Three
1 1 4 a
2 2 5 b
7 7 1 f
9 9 1 g
> dfMultiple
One Two Three
3 3 3 c
4 4 6 d
5 5 2 d
6 6 7 e
8 8 8 e
10 10 9 c
I tried with the duplicated() function
dfSingle = subset(df, !duplicated(df$Three))
dfMultiple = subset(df, duplicated(df$Three))
but it does not work as the first of the "c", "d" and "e" go to the "dfSingle".
I also tried to do a for-loop
MulipleValues = unique(df$Three[c(which(duplicated(df$Three)))])
dfSingle = data.frame()
x = 1
dfMultiple = data.frame()
y = 1
for (i in 1:length(df$One)) {
if(df$Three[i] %in% MulipleValues){
dfMultiple[x,] = df[i,]
x = x+1
} else {
dfSingle[y,] = df[i,]
y = y+1
}
}
It seems to do the right thing as the data frames have now the right amont of rows but they somehow have 0 columns.
> dfSingle
data frame with 0 columns and 4 rows
> dfMultiple
data frame with 0 columns and 6 rows
What am I doing wrong? Or is there another way to do this?
Thanks for your help!

Answer 1

In base R, we can use split with duplicated which will return you list of two dataframes.
df1 <- split(df, duplicated(df$Three) | duplicated(df$Three, fromLast = TRUE))
df1
#$`FALSE`
# One Two Three
#1 1 4 a
#2 2 5 b
#7 7 1 f
#9 9 1 g
#$`TRUE`
# One Two Three
#3 3 3 c
#4 4 6 d
#5 5 2 d
#6 6 7 e
#8 8 8 e
#10 10 9 c
where df1[[1]] can be considered as dfSingle and df1[[2]] as dfMultiple.

Answer 2

Here is a dplyr one for fun,
library(dplyr)
df %>%
group_by(Three) %>%
mutate(new = n() > 1) %>%
split(.$new)
which gives,
$`FALSE`
# A tibble: 4 x 4
# Groups: Three [4]
One Two Three new
<dbl> <dbl> <fct> <lgl>
1 1 4 a FALSE
2 2 5 b FALSE
3 7 1 f FALSE
4 9 1 g FALSE
$`TRUE`
# A tibble: 6 x 4
# Groups: Three [3]
One Two Three new
<dbl> <dbl> <fct> <lgl>
1 3 3 c TRUE
2 4 6 d TRUE
3 5 2 d TRUE
4 6 7 e TRUE
5 8 8 e TRUE
6 10 9 c TRUE

Answer 3

A way with dplyr:
library(dplyr)
df %>%
group_split(Duplicated = (add_count(., Three) %>% pull(n)) > 1)
Output:
[[1]]
# A tibble: 4 x 4
One Two Three Duplicated
<dbl> <dbl> <fct> <lgl>
1 1 4 a FALSE
2 2 5 b FALSE
3 7 1 f FALSE
4 9 1 g FALSE
[[2]]
# A tibble: 6 x 4
One Two Three Duplicated
<dbl> <dbl> <fct> <lgl>
1 3 3 c TRUE
2 4 6 d TRUE
3 5 2 d TRUE
4 6 7 e TRUE
5 8 8 e TRUE
6 10 9 c TRUE

Answer 4

You can do it using base R
One = c(1,2,3,4,5,6,7,8,9,10)
Two = c(4,5,3,6,2,7,1,8,1,9)
Three = c("a", "b", "c", "d","d","e","f","e","g","c")
df <- data.frame(One, Two, Three)
str(df)
df$Three <- as.character(df$Three)
df$count <- as.numeric(ave(df$Three,df$Three,FUN = length))
dfSingle = subset(df,df$count == 1)
dfMultiple = subset(df,df$count > 1)