Question

I have a simulation dataset that explores a set of parameter space, and each set of parameter are run multiple times (iterations), it looks like so:

p1    p2    p3  iteration  result
=================================
v3    v2    v1      1       23.8
v2    v1    v3      2       20.36
v3    v2    v1      2       28.8
v2    v1    v3      1       29.36
...

As can be seen from this example, both (v3, v2, v1) and (v2, v1, v3) are run twice. I am trying to extract only the rows with max result for each parameter setting, in this example:
only row 3 and 4 should be kept, as they represent the best results from that parameter set. Is there a easy way to accomplish that in R? Thanks

Answer 1

df <- read.table(textConnection("p1    p2    p3  iteration  result
v3    v2    v1      1       23.8
v2    v1    v3      2       20.36
v3    v2    v1      2       28.8
v2    v1    v3      1       29.36"), header = T)

library(plyr)
ddply(df, .(p1,p2,p3), function(x) return(x[(which(x$result == max(x$result))), ]))

p1 p2 p3 iteration result
1 v2 v1 v3         1  29.36
2 v3 v2 v1         2  28.80