从小标题的第1行减去第2行

Question

I have a tibble object in R, which looks like the following:

For each column, I want to subtract row1 from row2. My current approach is below:

mean.diff <- each.group.mean[1,2:num.groups]-each.group.mean[2,2:num.centroids]

What is confusing is that to execute this statement on the 2x9660 tibble, the code takes 8-15 secs, as determined by a Sys.time() calculation. Would appreciate any help to make this faster and to understand why this command takes so long.

Answer 1

By unlist ing, we could convert this to a vector and make it more efficient

unlist(each.group.mean[1,2:num.groups], use.names = FALSE)-
    unlist(each.group.mean[2,2:num.centroids], use.names = FALSE)

The reason is that a data.frame or tibble etc have many class attributes that makes it slower to do the subset and then the computation. By converting to vector all those classes are stripped off. Using a reproducible example

data

set.seed(24)
df1 <- as.tibble(as.data.frame(matrix(sample(1:10, 9660*10, 
     replace = TRUE), ncol = 9660)))

Benchmarks

system.time(df1[1,] - df1[2,])
#   user  system elapsed 
#   0.78    0.00    0.78 

system.time(unlist(df1[1,], use.names = FALSE) - unlist(df1[2,], use.names = FALSE))
#   user  system elapsed 
#   0.03    0.00    0.03