How to apply a custom recursive function with data.table and loop over each index group-wise?-CodePudding

Since I can't find an answer in below questions:

CodePudding user response：

Does this use of Reduce do the trick?

tmp = data.table(
  grp = c(rep(0,6), rep(1,6)),
  x=c(10,20,30,40,50,60,1,2,3,4,5,6),
  y=c(1,2,3,4,5,6, 10,20,30,40,50,60)
)
tmp[, z:=Reduce(f=function(z,i) z   x[i-1] - y[i-1],
                x=(1:.N)[-1],
                init=0,
                accumulate = T)
    ,by=grp
]

Output:

    grp  x  y    z
 1:   0 10  1    0
 2:   0 20  2    9
 3:   0 30  3   27
 4:   0 40  4   54
 5:   0 50  5   90
 6:   0 60  6  135
 7:   1  1 10    0
 8:   1  2 20   -9
 9:   1  3 30  -27
10:   1  4 40  -54
11:   1  5 50  -90
12:   1  6 60 -135

Take for example, row 4. The value in the z column is 54, which is equal to the prior row's z-value prior row's x-value, minus prior row's y-value.