dummyVars: Create A Full Set of Dummy Variables

Description

dummyVars creates a full set of dummy variables (i.e. less than full rank parameterization)

Usage

dummyVars(formula, ...)
# S3 method for default
dummyVars(formula, data, sep = ".", levelsOnly = FALSE, fullRank = FALSE, ...)
# S3 method for dummyVars
print(x, ...)
# S3 method for dummyVars
predict(object, newdata, na.action = na.pass, ...)
contr.ltfr(n, contrasts = TRUE, sparse = FALSE)
class2ind(x, drop2nd = FALSE)

Value

The output of dummyVars is a list of class 'dummyVars' with elements

call: the function call
form: the model formula
vars: names of all the variables in the model
facVars: names of all the factor variables in the model
lvls: levels of any factor variables
sep: NULL or a character separator
terms: the terms.formula object
levelsOnly: a logical

The predict function produces a data frame.

class2ind returns a matrix (or a vector if drop2nd = TRUE).

contr.ltfr generates a design matrix.

Arguments

formula: An appropriate R model formula, see References
...: additional arguments to be passed to other methods
data: A data frame with the predictors of interest
sep: An optional separator between factor variable names and their levels. Use sep = NULL for no separator (i.e. normal behavior of model.matrix as shown in the Details section)
levelsOnly: A logical; TRUE means to completely remove the variable names from the column names
fullRank: A logical; should a full rank or less than full rank parameterization be used? If TRUE, factors are encoded to be consistent with model.matrix and the resulting there are no linear dependencies induced between the columns.
x: A factor vector.
object: An object of class dummyVars
newdata: A data frame with the required columns
na.action: A function determining what should be done with missing values in newdata. The default is to predict NA.
n: A vector of levels for a factor, or the number of levels.
contrasts: A logical indicating whether contrasts should be computed.
sparse: A logical indicating if the result should be sparse.
drop2nd: A logical: if the factor has two levels, should a single binary vector be returned?

Author

contr.ltfr is a small modification of contr.treatment by Max Kuhn

Details

Most of the contrasts functions in R produce full rank parameterizations of the predictor data. For example, contr.treatment creates a reference cell in the data and defines dummy variables for all factor levels except those in the reference cell. For example, if a factor with 5 levels is used in a model formula alone, contr.treatment creates columns for the intercept and all the factor levels except the first level of the factor. For the data in the Example section below, this would produce:

 (Intercept) dayTue dayWed dayThu dayFri daySat daySun
           1      0      0      0      0      0      0
           1      0      0      0      0      0      0
           1      0      0      0      0      0      0
           1      0      1      0      0      0      0
           1      0      1      0      0      0      0
           1      0      0      0      1      0      0
           1      0      0      0      0      1      0
           1      0      0      0      0      1      0
           1      0      0      0      1      0      0

In some situations, there may be a need for dummy variables for all the levels of the factor. For the same example:

 dayMon dayTue dayWed dayThu dayFri daySat daySun
      1      0      0      0      0      0      0
      1      0      0      0      0      0      0
      1      0      0      0      0      0      0
      0      0      1      0      0      0      0
      0      0      1      0      0      0      0
      0      0      0      0      1      0      0
      0      0      0      0      0      1      0
      0      0      0      0      0      1      0
      0      0      0      0      1      0      0

Given a formula and initial data set, the class dummyVars gathers all the information needed to produce a full set of dummy variables for any data set. It uses contr.ltfr as the base function to do this.

class2ind is most useful for converting a factor outcome vector to a matrix (or vector) of dummy variables.

References

https://cran.r-project.org/doc/manuals/R-intro.html#Formulae-for-statistical-models

Examples

Run this code

when <- data.frame(time = c("afternoon", "night", "afternoon",
                            "morning", "morning", "morning",
                            "morning", "afternoon", "afternoon"),
                   day = c("Mon", "Mon", "Mon",
                           "Wed", "Wed", "Fri",
                           "Sat", "Sat", "Fri"),
                           stringsAsFactors = TRUE)

levels(when$time) <- list(morning="morning",
                          afternoon="afternoon",
                          night="night")
levels(when$day) <- list(Mon="Mon", Tue="Tue", Wed="Wed", Thu="Thu",
                         Fri="Fri", Sat="Sat", Sun="Sun")

## Default behavior:
model.matrix(~day, when)

mainEffects <- dummyVars(~ day + time, data = when)
mainEffects
predict(mainEffects, when[1:3,])

when2 <- when
when2[1, 1] <- NA
predict(mainEffects, when2[1:3,])
predict(mainEffects, when2[1:3,], na.action = na.omit)


interactionModel <- dummyVars(~ day + time + day:time,
                              data = when,
                              sep = ".")
predict(interactionModel, when[1:3,])

noNames <- dummyVars(~ day + time + day:time,
                     data = when,
                     levelsOnly = TRUE)
predict(noNames, when)

head(class2ind(iris$Species))

two_levels <- factor(rep(letters[1:2], each = 5))
class2ind(two_levels)
class2ind(two_levels, drop2nd = TRUE)

Run the code above in your browser using DataLab