How to Find the Difference Between Two Vectors in R

R's base setdiff() covers most vector-difference needs in one call, but like every set-based approach it drops duplicates and direction matters. Here are the four approaches worth knowing, including the dplyr equivalent for data frames.

setdiff(): the standard base R function

a <- c("apple", "banana", "cherry", "banana")
b <- c("banana", "date")

a_only <- setdiff(a, b)
print(a_only)  # "apple"  "cherry" - duplicate "banana" dropped, order not guaranteed

setdiff(a, b) treats both vectors as sets: any duplicate values in a collapse to one occurrence in the output, and the direction matters - setdiff(b, a) would give you a completely different result (just "date" here).

a[!a %in% b]: preserving duplicates and order

a <- c("apple", "banana", "cherry", "banana")
b <- c("banana", "date")

a_only <- a[!a %in% b]
print(a_only)  # "apple"  "cherry" - both "banana"s filtered, "apple" and "cherry" keep their position

%in% checks each element of a against the whole of b and returns a logical vector, which ! then negates to keep only non-matches. Unlike setdiff(), this is a filter rather than a set operation - it preserves a's original order and any duplicate values that aren't themselves in b.

Symmetric difference: items unique to either vector

a <- c("apple", "banana", "cherry")
b <- c("banana", "date")

sym_diff <- union(setdiff(a, b), setdiff(b, a))
print(sym_diff)  # "apple"   "cherry"  "date"

R has no single function for symmetric difference, but combining both one-way setdiff() calls with union() gets there in one line - everything unique to a plus everything unique to b.

dplyr anti_join(): the data frame equivalent

library(dplyr)

df_a <- data.frame(id = c(1, 2, 3), name = c("a", "b", "c"))
df_b <- data.frame(id = c(2, 3, 4), name = c("b", "c", "d"))

only_in_a <- anti_join(df_a, df_b, by = "id")
print(only_in_a)  # row with id = 1

When you're comparing data frames rather than plain vectors and want to match rows by a key column, anti_join() is the tidyverse equivalent of a vector difference - it returns rows from df_a whose id doesn't appear in df_b, without requiring the two frames to have identical columns the way setdiff() on whole data frames would.

NA handling: the pitfall that catches people out

NA in R is never equal to another NA under standard equality rules, but %in% and setdiff() specifically treat NA as matching NA - which is the opposite of what == would do. If your vectors contain NA values and you need standard IEEE-style behavior instead, filter them out explicitly with na.omit() before comparing.