R subsetting rows where values in multiple columns don't match - r

Question

Apologies if this has been asked already, but I searched and could not find an exact example of what I am trying to do. I'm trying to subset a dataframe to exclude rows that have matching numerical values across five columns. For example, for the following dataframe, df, I'd want to return a new dataframe only with rows 1:2, 5:6, and 8:10:
Row A B C D E
1 1 1 2 3 1
2 4 1 2 3 5
3 2 2 2 2 2
4 5 5 5 5 5
5 4 4 2 3 4
6 2 1 3 5 2
7 3 3 3 3 3
8 3 2 5 3 3
9 2 1 2 2 4
10 3 3 3 2 3
I'm having trouble figuring out how to do this for more than two columns. I've tried the following and know they are not right.
df2 <- df[!duplicated(df, c("A", "B", "C", "D", "E"))]
and
df2 <- df[df$A==df$B==df$C==df$D==df$E,]
Thanks in advance.

Answer 1

Data frames are usually operated on column-wise rather than row-wise, which is why your duplicated attempt doesn't work. (It's checking for duplicate rows within those columns.) And your == doesn't work because == is a binary operator, df$A == df$B will be TRUE or FALSE, and then (df$A == df$B) == df$C (implied parentheses) will be testing if df$C is TRUE or FALSE.
apply is a good way to run a function on each row. It will convert your data frame to a matrix to run the function, but in this case that's fine columns A through E are all numeric. Here's one way:
df[apply(df[, -1], 1, function(x) length(unique(x))) > 1, ]
# Row A B C D E
# 1 1 1 1 2 3 1
# 2 2 4 1 2 3 5
# 5 5 4 4 2 3 4
# 6 6 2 1 3 5 2
# 8 8 3 2 5 3 3
# 9 9 2 1 2 2 4
# 10 10 3 3 3 2 3
You could come up with all sorts of different functions to apply to test for all the elements being the same.
I assumed you actually have a column named Row. If that isn't the case, leave out the -1 in my code above.
Using this data, reproducibly shared with dput().
df = structure(list(Row = 1:10, A = c(1L, 4L, 2L, 5L, 4L, 2L, 3L,
3L, 2L, 3L), B = c(1L, 1L, 2L, 5L, 4L, 1L, 3L, 2L, 1L, 3L), C = c(2L,
2L, 2L, 5L, 2L, 3L, 3L, 5L, 2L, 3L), D = c(3L, 3L, 2L, 5L, 3L,
5L, 3L, 3L, 2L, 2L), E = c(1L, 5L, 2L, 5L, 4L, 2L, 3L, 3L, 4L,
3L)), .Names = c("Row", "A", "B", "C", "D", "E"), class = "data.frame", row.names = c(NA,
-10L))

Answer 2

You can simply compare all the columns against a single column and see if all the same
df[rowSums(df[-1] == df[, 1]) < (ncol(df) - 1), ]
# A B C D E
# 1 1 1 2 3 1
# 2 4 1 2 3 5
# 5 4 4 2 3 4
# 6 2 1 3 5 2
# 8 3 2 5 3 3
# 9 2 1 2 2 4
# 10 3 3 3 2 3
Or just df[rowSums(df == df[, 1]) < (ncol(df)), ]
Or similarly, you can avoid matrix conversions all together and combine Reduce and lapply
df[!Reduce("&" , lapply(df, `==`, df[, 1])), ]
# A B C D E
# 1 1 1 2 3 1
# 2 4 1 2 3 5
# 5 4 4 2 3 4
# 6 2 1 3 5 2
# 8 3 2 5 3 3
# 9 2 1 2 2 4
# 10 3 3 3 2 3