1

From a data frame of many columns, I would like to aggregate (i.e. sum) hundreds of columns by a single column, without specifying each of the column names.

Some sample data:

names <- floor(runif(20, 1, 5))
sample <- cbind(names)

for(i in 1:20){
col <- rnorm(20,2,4)
sample <- cbind(sample, col)
}

What I have until now is the following code, but it gives me that arguments must be the same length.

aggregated <- aggregate.data.frame(sample[,c(2:20)], by = as.list(names),     FUN = 'sum')

Original dataset is a lot bigger, so I can't specify the name of each of the columns to be aggregated and I can't use the list function.

M--
  • 25,431
  • 8
  • 61
  • 93
Santiago Veintimilla
  • 41
  • 1
  • 5

2 Answers2

1

You don't actually need to list them at all:

aggregate(. ~ names, sample, sum) # . represents all other columns

Of course base R is my favorite but in case someone wants dplyr:

library(dplyr)
data.frame(sample) %>% 
             group_by(names) %>% 
             summarise_each(funs(sum))
M--
  • 25,431
  • 8
  • 61
  • 93
0

Just alter your code slightly:

aggregated <- aggregate(sample[,c(2:20)], by = list(names), FUN = 'sum')

Mr. Bugle
  • 325
  • 2
  • 12