passing variable names to dplyr

[Roger Bos]

All,

I would like to figure out how to pass variable names to the dplyr function mutate.  For example, this works because hp is one of the variable names on mtcars:

mutate(mtcars, scale(hp))

Let's says I want to pass in the target variable instead of hard-coding the name, as follows:

target <- "hp"

mutate(mtcars, scale(target))

That dones't work.  I read somewhere about using lapply, but that suggestion didn't work for me either:

target <- lapply("hp", as.symbol)

mutate(mtcars, scale(target))

Does anyone know how to do this?

P.S.  I originally posted this on R-help.  I normally would not cross-post, but someone suggested that this was a better place for this question.

Thanks,

Roger

[Hadley Wickham]

Hi Roger,

There's currently no built-in support, although you can use existing R
tools to generate a call:

call <- substitute(mutate(mtcars, scale(target)), list(target =
as.name(target)))
eval(call)

(more info at http://adv-r.had.co.nz/Computing-on-the-language.html)

Figuring out how to express problem like this elegantly (and
efficiently) is on my long-term to do list for dplyr.

Hadley

> --
> You received this message because you are subscribed to the Google Groups
> "manipulatr" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to manipulatr+...@googlegroups.com.
> To post to this group, send email to manip...@googlegroups.com.
> Visit this group at http://groups.google.com/group/manipulatr.
> For more options, visit https://groups.google.com/groups/opt_out.



--
http://had.co.nz/

[Roger Bos]

Hadley,

Thanks for the quick reply and the suggestion about substitute.  It works fine for my needs.

Thanks so much for the dplyr package.

Roger

[Ryan Kelly]

Thank you! this works great for now. 

On Monday, January 27, 2014 2:55:44 PM UTC-5, Hadley Wickham wrote:

[Christopher Wright]

I wonder if this is the same issue (still coming to grips with the scoping / eval rules in R - starting to read The-Book!

I have a dataframe (main_schools), with one column named "school", and a number of other columns. This dataframe is defined at the top level.

I want to summarise by school, and a number of other columns, so I thought I would write a function:

test <- function(col_name) {
    summarise(group_by(main_schools, school, col_name), freq = n())
}

and then pass in many <col_name> values

but that doesn't work :

> test(facilities)
 Error: unknown column 'col_name'

I expected (wrongly - and I'm trying to educate myself), that <col_name> would be evaluated in the environment of the function, and the value <safe> be found.

So is my alternative to just copy/paste
    summarise(group_by(main_schools, school, col_name), freq = n())

with a different value for "col_name" each time, like:

    summarise(group_by(main_schools, school, facilities), freq = n())
    summarise(group_by(main_schools, school, transport), freq = n())

etc etc?

dplyr is a wonderful package, and I'm very grateful!

with thanks

Chris

[Doug Mitarotonda]

Read up on the “_” versions of all of the dplyr functions to understand how to program on the language. 

For example (note how most people who use dplyr use the %>% function instead of the nesting you used):

dft <- data_frame(A = sample(1:3, 10, TRUE), B = sample(1:10, 10, TRUE))

col_name <- "A"

dft %>% group_by_(col_name) %>% summarize(freq = n())

Source: local data frame [3 x 2]

  A freq

1 1    2

2 2    4

3 3    4

--
You received this message because you are subscribed to the Google Groups "manipulatr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to manipulatr+...@googlegroups.com.
To post to this group, send email to manip...@googlegroups.com.
Visit this group at http://groups.google.com/group/manipulatr.

For more options, visit https://groups.google.com/d/optout.

[Christopher Wright]

Thanks Doug,

So now we seem to have another scoping / laziness issues:

> test <- function(col_name) {
+     summarise(group_by_(main_schools, school, col_name), freq = n())
+ }
> test("site")

 Error in as.lazy_dots(list(...)) : object 'school' not found

I'm off to read about evaluation (lazy evaluation ! ) and scoping in R!

Chris

[Doug Mitarotonda]

Consider what the _ does, it switches *all arguments* to standard evaluation. You can’t mix and match standard and non-standard evaluation (how would R know which to do for each argument?). If you put “school”, i.e., as a string, your code should work.

> dft <- data_frame(A = sample(1:3, 10, TRUE), B = sample(1:10, 10, TRUE), C = sample(4:6, 10, TRUE))

> col_name <- "A"

> dft %>% group_by_(col_name, "B") %>% summarize(freq = n())

Source: local data frame [9 x 3]

Groups: A

  A  B freq

1 1  2    1

2 1  6    1

3 1  9    1

4 2  1    1

5 2  2    1

6 2  7    1

7 2 10    1

8 3  4    1

9 3  5    2