David Z David Z - 21 days ago 9
R Question

How to get the cumulative sum by group in R?

Suppose I have a dataframe such that:

df<-data.frame(id=1:8,group=c(1,0,0,1,1,0,1,0),rep=c(rep("d1",4),rep("d2",4)),value=rbinom(8,1,0.6))
df
id group rep value
1 1 1 d1 0
2 2 0 d1 0
3 3 0 d1 0
4 4 1 d1 1
5 5 1 d2 1
6 6 0 d2 0
7 7 1 d2 1
8 8 0 d2 1


What's the best way to get the cumulative sum by
group
and
rep
such that:

cumsum
group d1 d1+d2 d1+d2+d3
0 0 1 ...
1 1 3 ...

Answer Source
library(data.table)

# convert to data.table in place
setDT(df)

# dcast and do individual sums
dt.cast = dcast.data.table(df, group ~ rep, value.var = 'value',
                           fun.aggregate = sum)
dt.cast
#   group d1 d2
#1:     0  0  1
#2:     1  1  2

# cumsum
dt.cast[, as.list(cumsum(unlist(.SD))), by = group]
#   group d1 d2
#1:     0  0  1
#2:     1  1  3