Calculate relative cumulative sum in R

Question

My sample data looks like this:

data <- read.table(text="group;  year;    val
               a;        1928;    20
               a;        1929;    50
               a;        1930;    40
               a;        1931;    45
               b;        1935;   -10
               b;        1936;   -15 ", sep=";", header=T, stringsAsFactors = FALSE)

> data
                 group year val
1                    a 1928  20
2                    a 1929  50
3                    a 1930  40
4                    a 1931  45
5                    b 1935 -10
6                    b 1936 -15

What I would like to do is to calculate the cumulative sum relative to 1930 in a new column sum_rel (eg 1930 is the start year, all values over 1930 should be added and under 1930 should be subtracted). If all years are bigger than 1930 the relative value (0=) should be the lowest year per group (as in case b).

               group year  val sum_rel
               a     1927      -110
               a     1928  20  -90
               a     1929  50  -40
               a     1930  40    0
               a     1931  45   45
               b     1934      0
               b     1935 -10  -10
               b     1936 -15  -25

I had a look at the cumsum function but couldn't figure out to apply it over groups and I would be very glad if you could help me.

Answer 1

Adding the extra row is probably the trickiest part. This seems to produce the output you are after

do.call("rbind", unname(lapply(split(data, data$group), function(x) {
    x<-x[order(x$year),]
    cx <- c(which(x$year==1930),0)[1]+1
    cs <- cumsum(c(0, x$val))
    cbind(rbind(transform(x[1,], val=NA, year=min(x$year)-1), x), sum_rel=cs-cs[cx])
})))

#               group year val sum_rel
# 1                 a 1927  NA    -110
# 2                 a 1928  20     -90
# 3                 a 1929  50     -40
# 4                 a 1930  40       0
# 5                 a 1931  45      45
# 52                b 1934  NA       0
# 51                b 1935 -10     -10
# 6                 b 1936 -15     -25