data.table: cumulative values by grouping variable

Question

I have data

set.seed(42)
dat <- data.table(id=1:8, group=c(1,1,2,2,2,3,3,3), val=rnorm(8))

> dat
   id group         val
1:  1     1  1.37095845
2:  2     1 -0.56469817
3:  3     2  0.36312841
4:  4     2  0.63286260
5:  5     2  0.40426832
6:  6     3 -0.10612452
7:  7     3  1.51152200
8:  8     3 -0.09465904

and I would like to obtain the cumulative values of val within each level of group.

> res
   id group         cum
1:  1     1  1.37095845
2:  2     1  0.80626037
3:  3     2  0.36312841
4:  4     2  0.995991
5:  5     2  1.400259
6:  6     3 -0.10612452
7:  7     3  1.405397
8:  8     3  1.310738

I am always astonished by the efficiency of data.table, so I 'm wondering about a way to get this done in data.table but of course any other efficient solution is just as welcome.

What have you attempted so far? Do you know `?cumsum` – talat Feb 27 '18 at 16:33 — talat, Feb 27 '18 at 16:33

sm925 · Accepted Answer · 2018-02-27T16:41:33.990

0

You can do it using cumsum:

dat[, cum:= cumsum(val), by = group]
dat[, val := NULL]

   id group        cum
1:  1     1  1.3709584
2:  2     1  0.8062603
3:  3     2  0.3631284
4:  4     2  0.9959910
5:  5     2  1.4002593
6:  6     3 -0.1061245
7:  7     3  1.4053975
8:  8     3  1.3107384

edited Feb 27 '18 at 16:41

answered Feb 27 '18 at 16:34

sm925

2,648
1
16
28

data.table: cumulative values by grouping variable

1 Answers1

Linked