Revision history [back]

To utilize a vector containing column names as an input for dplyr::group_by(), you can use the {{ }} curly-curly syntax for non-standard evaluation (NSE) along with the !! bang-bang operator to interpolate the variable.

For example, if your vector containing column names is called "vars", you can use the following code:

library(dplyr)

vars <- c("col1", "col2", "col3")

df %>% 
  group_by({{vars}})

The curly-curly syntax {{ }} tells R to evaluate the expression inside the braces as a variable name rather than as a character string, while the bang-bang operator !! tells R to interpolate the variable.

Note that you can also use the standard evaluation syntax by converting your vector to a list using the function as.list(), as follows:

vars <- c("col1", "col2", "col3")

df %>% 
  group_by_at(vars = as.list(vars))

The function groupbyat() allows you to specify the columns to group by using a character vector of column names, a numeric vector of column positions, or a list of column specifications.