R: Conditional diff function

user_11572623 0
Viewed 197

I would like to calculate the consecutive differences between the rows of a variable of a dataframe in R, exactly like the diff() function, however, only between the rows with the same id number.

Dummy data:

id <- c(1,1,1,1,1,1,1,1,2,2,2,2,2,2,2)

t1 <- c(4,3,4,4,5,8,9,11,3,8,9,7,9,10,3)

df <- data.frame(id, t1)

Using diff(df$t1) I get:

-1 1 0 1 3 1 2 -8 5 1 -2 2 1 -7

I would like:

-1 1 0 1 3 1 NA -8 5 1 -2 2 1 -7

I tried also:

df%>%
  group_by(id)%>%
  diff(df$t1)

But I get the error:

Error in diff.default(., df$t1) : 'lag' and 'differences' must be integers >= 1

Any ideas?

diff r group-by
3 Answers

A one-liner base R could be

ave(df$t1, df$id, FUN = function(x) c(x[-1], NA) - x)
#[1] -1  1  0  1  3  1  2 NA  5  1 -2  2  1 -7 NA
user_82454060

You would need to move diff() to inside a mutate() statement if you are using dplyr. But diff() returns a vector that's shorter by 1 than your input vector which makes it difficult to keep the same number of rows. An alternative is to use the dplyr lead() function to grab the "next" value in the group

df%>%
  group_by(id)%>%
  mutate(diff=lead(t1)-t1)
user_23720640

An option with data.table

library(data.table)
setDT(df)[, Diff = shift(t1, type = 'lead') - t1, id]
user_37322710
Related