CODEWITHSUNDEEP
rlistparameter-passingaggregate
Julia Perl
Julia Perl

Reputation: 13

Aggregate in R with parameter list

I'm very new to R and running out of ideas how to solve the following problem :(

My dataset 'test' looks like this

A  B  C  y     z
a1 b1 c1 0.10  0
a1 b1 c2 0.01  1
a1 b2 c1 0.20  1
a1 b2 c2 0.10  0
a2 b1 c1 0.10  0
a2 b1 c2 0.01  1
a2 b2 c1 0.20  0
a2 b2 c2 0.30  1

I want to aggregate my dataset by the some of the 'y' values of the two dimensions 'A' and 'B' which can be done by

> aggregate(x = test$y, by = list(test$A, test$B), FUN=sum)

and returns the correct result:

Group.1 Group.2 x
   a1      b1   1
   a2      b1   1
   a1      b2   1
   a2      b2   1

So far, so good. In this simple case I can explicitly write the column names, but what if I want to parameterize them? Somehow like

 > fields = 'test$A, test$B'
 > aggregate(x = test$aL, by = list(.(fields)), FUN=sum)

it throws an error that the arguments must have the same length. So how can I parameterize the aggregate list? I would be very grateful for any tips.

Upvotes: 1

Views: 378

Answers (2)

akrun
akrun

Reputation: 887851

In addition to aggregate based options in the comments, the syntax for some efficient methods such as data.table or dplyr are below.

We convert the 'data.frame' to 'data.table' (setDT(test)), grouped by 'fields', get the sum of 'y'

library(data.table)
fields <- c("A", "B")
setDT(test)[, .(y = sum(y)), by = fields]
#    A  B    y
#1: a1 b1 0.11
#2: a1 b2 0.30
#3: a2 b1 0.11
#4: a2 b2 0.50

Or using dplyr, we can pass the objects in group_by_ with .dots argument and get the sum of 'y'.

library(dplyr)
test %>% 
    group_by_(.dots = fields) %>% 
    summarise(y = sum(y))
#    A     B     y
#  <chr> <chr> <dbl>
#1    a1    b1  0.11
#2    a1    b2  0.30
#3    a2    b1  0.11
#4    a2    b2  0.50

Upvotes: 1

Sotos
Sotos

Reputation: 51592

One way is to make fields a list with your variables i.e.

fields <- list(test$A, test$B)
aggregate(test$y, by = fields, FUN=sum)

or create a function,

fun1 <- function(v1, v2){aggregate(test$y, by = list(v1, v2), FUN = sum)}
fun1(test$A, test$B)

Upvotes: 1

Related Questions