从旧数据表构建新数据表

Question

I have a data table DT with columns k,v: (k1, v1), (k2, v2), (k3, v3)...

I would like to group values by key and then do something with them and produce a new data table (or frame) (s1, t1), (s2, t2)... .

Each grouped (k, v1, ..vi) set can be mapped to 0 or more rows in the destination data.table dest , which have totally different range of keys, not related to k.

I understand that I can write: DT[, myfun(k,v), by=k] but I am not sure how that helps me.

For example, assume that my function myfun needs to compute s = prod(v1, ...vi) and then to increase row s by k : dest[s] += k .

I am not sure how can I code this in R!

Answer 1

You create an summarised version of DT:

DT1 <- DT[, prod(v), by=k]

and then use the values of this to increase dest , the $v now equals the s in your post and so identifies the rows to be incremented:

dest[DT1$v, "columnToBeIncreased"] = dest[DT1$v, "columnToBeIncreased"] +DT1$k

You say dest is a data.table, and you want to increase a row by a given amount. If you want the whole row (ie all the columns in that row) to be increased, omit the "columnToBeIncreased" (remove even the quotes, but leave the preceeding comma) - otherwise set it to the the name of the column you want to be incremented.