R - within and ave using multiple inputs -

i have been using functions with , ave in following manner:

dt = within(dt,{sim1 = ave(sample, no, fun = sample_ret)}) 

this function takes data.table dt , segments variable no. after that, applies function sample_ret input sample , creates column sim1 in data.table.

is there way can make function take multiple inputs? example, instead of taking sample, how can let function sample_ret take column along sample?

if not possible, there function can accomplish want? basically, want segment data.table , apply function using more 1 column inputs?

for example, data.table dt:

col1 col2 type col4 1    2       4 5    6       8 9    10   b   12 13   14   b   16 3    34     16 1    2    c   16 

and want segment dt type , how many values greater 6 in col1 , how many values greater 12 in col2 per each type. let's have function greater_than takes in 2 columns , that. how can accomplish want ave or other alternative?

a base r option be

indx <- df1[1:2] >val[col(df1[1:2])] rownames(indx) <- df1$type rowsum(indx+0l, row.names(indx)) #    col1 col2 #a    0    1 #b    2    1 #c    0    0 

or

library(dplyr) df1 %>%      group_by(type) %>%      mutate_each(funs(.>df2$.), col1:col2) %>%      summarise_each(funs(sum), col1:col2) #  type col1 col2 #1       0    1 #2    b    2    1 #3    c    0    0 

or possible data.table option suggested @david arenburg

library(data.table) cols <- paste0("col", 1:2) sums <- function(x, y) sum(x > y)  setdt(df1)[, map(sums, .sd, val), by=type, .sdcols = cols] #    type col1 col2 #1:       0    1 #2:    b    2    1 #3:    c    0    0 

data

df1 <- structure(list(col1 = c(1l, 5l, 9l, 13l, 3l, 1l), col2 = c(2l,  6l, 10l, 14l, 34l, 2l), type = c("a", "a", "b", "b", "a", "c" ), col4 = c(4l, 8l, 12l, 16l, 16l, 16l)), .names = c("col1",  "col2", "type", "col4"), class = "data.frame", row.names = c(na,  -6l))  df2 <- data.frame(col1=6, col2= 12) val <- c(6,12)