Compare two vectors (including multiple items) - r

Question

I need to find which elements in a new vector (vb) have been added to another vector (va). If there for example is unly one "2" in va, but two "2" in vb, then one "2" has been added.
The comment in the code below shows what is sought.
va <- c(1, 2) # Original vector
vb <- c(1, 2) # NA or NULL
vb <- c(2, 2) # 2
vb <- c(1, 1) # 1
vb <- c(1) # NA or NULL
vb <- c(2) # NA or NULL
vb <- c(3, 3) # c(3, 3)
I've tried match, union, intersect, %in%, etc. but can't get it to work to consider also multiple instances. This feels irritatingly simple...

Answer 1

The following reproduces your expected outcome. Just as an honest heads-up, I'm not really happy with my solution, this seems oddly convoluted:
f <- function(a, b) {
a <- as.data.frame(unclass(rle(a)));
b <- as.data.frame(unclass(rle(b)));
t <- merge(a, b, by = "values", all = TRUE);
t$lengths.x[is.na(t$lengths.x)] <- 0;
t$diff <- t$lengths.y - t$lengths.x;
t <- t[!is.na(t$diff) & t$diff > 0, ];
return(rep(t$values, t$diff));
}
va <- c(1, 2);
vb <- c(1, 2) # NA or NULL
f(va, vb);
#numeric(0)
vb <- c(2, 2) # 2
f(va, vb);
#[1] 2
vb <- c(1, 1) # 1
f(va, vb);
#[1] 1
vb <- c(1) # NA or NULL
f(va, vb);
#numeric(0)
vb <- c(2) # NA or NULL
f(va, vb);
#numeric(0)
vb <- c(3, 3) # c(3, 3)
#[1] 3 3
Explanation: I'm making use of rle to compare the lengths (level of duplicity) of different entries in va and vb; then report only those that are not already in va.
Update
Here is a much cleaner method using a recursive function.
f <- function(a, b) {
if (length(a) == 0 | length(b) == 0) return(NULL);
m <- data.frame(idx.a = 1:length(a), idx.b = match(a, b));
m <- m[complete.cases(m), ];
# Here is the recursive call
if (nrow(m) > 0) f(a[-m$idx.a[1]], b[-m$idx.b[1]]) else b;
}
va <- c(1, 2) # Original vector
f(va, c(1, 2));
#NULL
f(va, c(2, 2));
#[1] 2
f(va, c(1, 1));
#[1] 1
f(va, c(1));
#NULL
f(va, c(2));
#NULL
f(va, c(3, 3));
#[1] 3 3

Answer 2

Not the most elegant, but it works for all your cases:
Diff_frequency <- function(va,vb){
df <- merge(as.data.frame(table(va)), as.data.frame(table(vb)), by.x="va", by.y="vb", all=T)
df$Freq.x[is.na(df$Freq.x)] <- 0
df$Dif <- df$Freq.y - df$Freq.x
df$Dif[is.na(df$Dif) | df$Dif < 0] <- 0
return(rep(as.numeric(as.character(df[,1])), df$Dif))
}
Diff_frequency(va,vb)
Examples of output:
va=c(1,1,1,2,2,2,3)
vb=c(1,1,4,4,2,2,5)
Diff_frequency(va,vb)
[1] 4 4 5
va=c(1,1,1,2,2,2,3)
vb=c(1,1,1,1,2,2,2,3,3,5)
Diff_frequency(va,vb)
1] 1 3 5
va=c(1,1,1,2,2,2,3)
vb=c(1,1,2,3)
Diff_frequency(va,vb)
numeric(0)