dummies (version 1.5.4)

dummy: Flexible, efficient creation of dummy variables.

Description

This package flexibly and efficiently creates dummy variables for a variety of structures.

Usage

dummy(x, data = NULL, sep = "", drop = TRUE, fun = as.integer, verbose = FALSE)
dummy.data.frame(data, names = NULL, omit.constants=TRUE, dummy.classes = getOption("dummy.classes"), all = TRUE, ...)

Arguments

x

a single variable or variable _name_

data

an object such as a data.frame or matrix that has colnames

drop

Whether to drop (i.e. omit) dummy variables for unused levels. When x or data[,x] is a factor, this parameter variables for only the used levels. By default, dummies are created only for the used levels, i.e. TRUE

sep

For the names of the created dummy variables, sep is the character used between the variable name and the value.

fun

Function used to coerce values in the resulting matrix or frame.

verbose

logical. Whether to print(cat) the number of dummy variables created Default: FALSE

names

The names of the columns to expand to dummy variables. Takes precedent over dummy.classes parameter.

dummy.classes

( For dummy.data.frame only ) A vector of classes names for which dummy variables are created -or- "ALL" to create dummy variables for all columns irregardless of type. By default, dummy variables are produced for factor and cha

omit.constants

Whether to omit dummy variables that are constants, i.e. contain only one value. Overridden by drop==FALSE.

all

( For dummy.data.frame only ). Whether to return columns that are not dummy classes. The default is TRUE and returns all classes. Non dummy classes are untouched.

...

arguments passed to dummy

Value

• dummy returns a matrix with the number of rows equal to the that of given variable. By default, the matrix contains integers, but the exact type can be affected by fun argument. Rownames are retained if the supplied variable has associate row names.

  dummy.data.frame returns a data.frame in which variables are expanded to dummy variables if they are one of the dummy classes. The columns are return in the same order as the input with dummy variable columns replacing the original column.

Details

dummy take a single variable OR the name of single variable and a data frame. It coerces the variable to a factor and returns a matrix of dummy variables using model.matrix. If the data has rownames, these are retained.

Optionally, the parameter drop indicates that that dummy variables will be created for only the expressed levels of factors. Setting it to false will produce dummy variables for all levels of all factors.

If there is only one level for the variable and verbose == TRUE, a warning is issued before creating the dummy variable. Each element of this dummy variable, will have the same value.

A seperator, sep, can be specified for the seperator between the variable name and the value for the construction of new variable names. The default is to provide no seperator.

The type of values returned can be affected using the fun argument. fun is called on each of the resultant dummy variables. The only useful functions that the author has employed are as.interger (the default) or as.logical.

dummy.data.frame takes a data.frame or matrix and returns a data.frame in which all specified columns are expanded as dummy variables. Specific columns can be named with the names argument or specified on a class basis by the dummy.classes argument. Specified names take precedent over classes. The default is to expand dummy variables for character and factor classes, and can be controlled globally by options('dummy.classes').

If the argument all is FALSE. The resulting data.frame will contain only the new dummy variables. By default, all columns of the object are returned in the order of the original frame. Dummy variables are expanded in place.

omit.constants indicates whether to omit dummy variables that assume only a single value. This is the default. If drop==FALSE, constant variables are retained regardless of the setting.

References

http://wiki.r-project.org/rwiki/doku.php?id=tips:data-manip:create_indicator

http://tolstoy.newcastle.edu.au/R/help/00b/1199.html

http://tolstoy.newcastle.edu.au/R/help/03a/6409.html

http://tolstoy.newcastle.edu.au/R/help/01c/0580.html

Many other discussions on R-Help. Too many to list.

See Also

model.frame, model.matrix, factor

Examples

letters <- c( "a", "a", "b", "c", "d", "e", "f", "g", "h", "b", "b" )
  dummy( as.character(letters) )
  dummy( letters[1:6] )
  
  l <- as.factor(letters)[ c(1:3,1:6,4:6) ]
  dummy(l)
  dummy(l, drop=FALSE)
  dummy(l, sep=":")
  dummy(l, sep="::", fun=as.logical)
  
  # TESTING NAS
  l <- c( NA, l, NA)
  dummy(l)
  dummy(l,sep=":")
  
  
  dummy(iris$Species)
  dummy(iris$Species[ c(1:3,51:53,101:103) ] )
  dummy(iris$Species[ c(1:3,51:53,101:103) ], sep=":" )
  dummy(iris$Species[ c(1:3,51:53) ], sep=":", drop=FALSE )     
  

  # TESTING TRAP FOR ONE LEVEL
  dummy( as.factor(letters)[c(1,1,1,1)] )
  dummy( as.factor(letters)[c(1,1,2,2)] )
  dummy( as.factor(letters)[c(1,1,1,1)] , drop = FALSE )   

  
  dummy.data.frame(iris)
  dummy.data.frame(iris, all=FALSE)


  dummy.data.frame(iris, dummy.class="numeric" )
  dummy.data.frame(iris, dummy.class="ALL" )