create a filter expression (i) dynamically in data.table

Question

Having a data.table

library(data.table)
dd <- data.table(x=1:10,y=10:1,z=20:20)

I can filter it using

dd[x %in% c(1, 3) & z %in% c(12, 20)]
   x  y  z
1: 1 10 20
2: 3  8 20

Now I would like to create the same filter dynamically. This what I have tried so far:

cond <- list(x=c(1,3),z=c(12,20))
vars <- names(cond)
## dd[get(vars[[1]]) %in% cond[[1]] & get(vars[[2]]) %in% cond[[2]]]

EVAL = function(...){
  expr <- parse(text=paste0(...))
  print(expr)
  eval(expr)
  }

dd[ EVAL(vars, " %in% ", cond, collapse=" & ") ]

But I still get an error:

 Error in match(x, table, nomatch = 0L) : object 'x' not found

even if the expression evalutaed looks good:

expression(x %in% c(1, 3) & z %in% c(12, 20))

Is there a way to fix this?

You are probably looking for a general solution, but for your specific task, it seems to me that you could just do something like EVAL = function(x, vars, cond){ setkeyv(x, vars) ; x[do.call(CJ, cond), nomatch = 0L] } ; EVAL(dd, vars, cond) — David Arenburg
– David Arenburg, Commented Jan 24, 2016 at 0:01
@DavidArenburg yes I am looking for a general solution. But I like your solution. I think it is more efficient than my attempt to eavl/parse expressions.. can you please change your comment to an answer with some explanations? — agstudy
– agstudy, Commented Jan 24, 2016 at 0:13
dd[rowSums(mapply(`%in%`, dd[names(cond)], cond)) == length(cond), ] in base — rawr
– rawr, Commented Jan 24, 2016 at 1:22
The costly way is: take the Cartesian product of the elements of cond and then join the result: dd[do.call(CJ,cond), on=names(cond), nomatch=0] (Oh, I see David posted something similar above.) — Frank
– Frank, Commented Jan 24, 2016 at 20:24

jangorecki · Accepted Answer · 2016-01-24 01:09:30Z

5

Building expression instead of parsing it.

library(data.table)
dd = data.table(x=1:10,y=10:1,z=20:20)
AndIN = function(cond){
    Reduce(
        function(x, y) call("&", call("(",x), call("(",y)),
        lapply(names(cond), function(var) call("%in%", as.name(var), cond[[var]]))
    )
}
cond = list(x=c(1,3),z=c(12,20))
AndIN(cond)
#(x %in% c(1, 3)) & (z %in% c(12, 20))
dd[eval(AndIN(cond))]
#   x  y  z
#1: 1 10 20
#2: 3  8 20

Calls call("(",x) and call("(",y) may not be necessary.

edited Jan 24, 2016 at 1:09

answered Jan 24, 2016 at 0:59

jangorecki

16.8k5 gold badges86 silver badges169 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

create a filter expression (i) dynamically in data.table

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related