r - Conditional selection of variables -

suppose i've following table:

     var    iter_1  iter_2  iter_3  iter_4     var1    6       8       5       7     var2    5       1       7       8     var3    3       8       8       4     var4    8       7       2       5     var5    8       7       9       2     var6    8       7       3       6     var7    4       7       4       5 

i want select combinations of columns each row such there combinations equal specific sum. e.g. in case, suppose want each var combination of iter 15. in case want select var1, iter_2 & iter_4. var2, iter_3 & iter_4.

i wanted develop code such code can tell me column values select each var.

can plz suggest method? 1 don't need write code, logic can use.

thank you.

this works if sum taken columns:

data = data.frame(x = 1:3, y = 2:4, z = 5:7) sums = apply(data, 1, sum) target.val = 11 which(sums == target.val) 

otherwise looks exact cover problem. http://en.wikipedia.org/wiki/exact_cover

or

you use stochastic approach, genetic algorithm. simplistic solution:

find.colsums = function(data, target,  n.tries = 100) {   nrows = nrow(data)   max.cols = ncol(data)   n.columns = sample(max.cols, n.tries, replace = true)    (i in 1:n.tries){     test.cols = sample(max.cols, n.columns[i])      (row in 1:nrows){       if (sum(data[row, test.cols]) == target){         cat("match @ row:", row, "cols:", test.cols, "\n")       }     }   } } 

example:

data = data.frame(x = 1:3, y = 2:4, z = 5:7) target = 7 find.colsums(data, target) 

fun big dataset:

n = 1000 min.val = 1 max.val = 30 ncols = 10 target = ((min.val + max.val) * ncols/2)  data = matrix(sample(min.val:max.val, n, replace = true), ncol = ncols) find.colsums(data, target, n.tries = 1000)