Finding cumulative product if a certain condition is met in R

Question

I have a dataset that looks like the following (called Data):

v1 v2
1   1  
1   3
1   5
2   3
2   4
3   1
3   2

I want to return a vector v3 that:

• is equal to v2[i] if v1[i] is not equal to v1[i-1]
• is equal to v3[i-1]*v2[i] if v1[i] is equal to v1[i-1]

So, in this example, v3 should return

v3
1
3
15
3
12
1
2

I've lagged the column v1 by using lag.v1<-c(NA,Data[1:nrow(Data)-1,1]) in order to compare to the previous row. I think something similar to the following should work, but with the value of v3 in the previous row instead of the current row.

Data$v3<-ifelse(1*(Data$v1==lag.v1)==1, Data$v3*Data$v2, Data$v2)

In other words, I need to somehow access the previous row of v3 (lag v3) as I'm forming v3 in the above equation.

Help is greatly appreciated, thank you!

Answer 1

You can use ave with cumprod , this calculates the cumulative product of column v2 grouped by v1 :

df$v3 <- with(df, ave(v2, v1, FUN=cumprod))

df
#  v1 v2 v3
#1  1  1  1
#2  1  3  3
#3  1  5 15
#4  2  3  3
#5  2  4 12
#6  3  1  1
#7  3  2  2

---

With plyr package, you can use ddply with transform :

plyr::ddply(df, "v1", transform, v3 = cumprod(v2))

#  v1 v2 v3
#1  1  1  1
#2  1  3  3
#3  1  5 15
#4  2  3  3
#5  2  4 12
#6  3  1  1
#7  3  2  2

---

If you haven't, you probably also want to know a dplyr approach:

library(dplyr)
df %>% group_by(v1) %>% mutate(v3 = cumprod(v2))

#Source: local data frame [7 x 3]
#Groups: v1 [3]

#     v1    v2    v3
#  <int> <int> <dbl>
#1     1     1     1
#2     1     3     3
#3     1     5    15
#4     2     3     3
#5     2     4    12
#6     3     1     1
#7     3     2     2

Answer 2

We can use data.table

library(data.table)
setDT(df)[, v3 := cumprod(v2), by = v1]
df
#   v1 v2 v3
#1:  1  1  1
#2:  1  3  3
#3:  1  5 15
#4:  2  3  3
#5:  2  4 12
#6:  3  1  1
#7:  3  2  2