mungebits)

Description Usage Arguments Value See Also Examples

A mungebit which does not affect any existing columns, but can transform other columns into new columns, or affect multiple columns simultaneously as functions of each other.

1	multi_column_transformation(transformation)

transformation

a function. The arguments will be the ordered columns selected from the dataframe when the transformation is performed (see examples). One has to be careful to return an atomic vector if the result is 1-dimensional, and a list otherwise.

a function which takes a data.frame, input columns, and output columns, the latter two of which specify the domain and range of the transformation (see examples).

column_transformation

perimeter <- multi_column_transformation(function(x, y) 2*x + 2*y)
perimeter(iris, c('Sepal.Length', 'Sepal.Width'), 'Sepal.Perimeter')

multiplier <- multi_column_transformation(function(x, y) x*y)
multiplier(iris, c('Sepal.Length', 'Sepal.Width'), 'Sepal.Area')

property_generator <- multi_column_transformation(
   function(x, y) list(2*x + 2*y, x*y))
property_generator(iris, c('Sepal.Length', 'Sepal.Width'),
   c('Sepal.Perimeter', 'Sepal.Area'))

swapper <- multi_column_transformation(function(x, y) list(y, x))
swapper(iris, c('Sepal.Length', 'Sepal.Width'),
  c('Sepal.Width', 'Sepal.Length'))

column_remover <- multi_column_transformation(function(...) NULL)
column_remover(iris, c('Sepal.Length', 'Sepal.Width'))

scaling_fun <- function(...) {
  args <- list(...)
  const <- args[[length(args)]]
  args <- head(args, -1)
  if (length(args) == 1) args[[1]] * const
  else lapply(args, function(col) col * const)
}
scaler <- multi_column_transformation(scaling_fun)
# scale Sepal.Length and Sepal.Width by two
scaler(iris, c('Sepal.Length', 'Sepal.Width'), , 2)
scaler(iris[c('Petal.Length', 'Petal.Width')], , , 2)
# Note the missing second and third arguments.