Remove rows based on mode

Question

Data:

 df <- data.frame("y"= c(rep(100,10), rep(103,10)), "x" = c(rep(1,8),2,2,rep(4,5),7,7, rep(4,3)) )

For each value of y, we expect a unique value of x. But in data, we have a couple of unwanted values. Based on mode, we can find the corresponding value of x for each value of y.

Can we identify the row numbers where there is unwanted pairing ?

Expected output: 9, 10, 16, 17

please run the code . You can see x and y values. When y = 100, x is 1 (8 times) or 2( 2 times). So right pair is (y=100, x=1) so (y=100,x =2) is unwanted and is to be removed — Navin Manaswi, Apr 26 '16 at 07:07

score 3 · Answer 1 · edited May 23 '17 at 11:52

3

We can use ave to group by 'y', get the logical index of elements in 'x' that are not the 'Mode', from there get the row index with which.

with(df, which(as.logical(ave(x, y, FUN= function(x) x!=Mode(x)))))
#[1]  9 10 16 17

Or as @thelatemail mentioned in the comments, the above can be made compact

which(with(df, x != ave(x, y, FUN=Mode) ))

where Mode is (taken from here)

Mode <- function(x) {
  ux <- unique(x)
 ux[which.max(tabulate(match(x, ux)))]
}

edited May 23 '17 at 11:52

Community

1
1

answered Apr 26 '16 at 05:32

akrun

874,273
37
540
662

1

You could save a bunch of comparisons and do `which(with(df, x != ave(x, y, FUN=Mode) ))` – thelatemail Apr 26 '16 at 05:48
@thelatemail That is a nice compact approach. Thanks – akrun Apr 26 '16 at 05:50

Remove rows based on mode

1 Answers1