contr_one_hot.Rd

% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/contr_one_hot.R
\name{contr_one_hot}
\alias{contr_one_hot}
\title{Contrast function for one-hot encodings}
\usage{
contr_one_hot(n, contrasts = TRUE, sparse = FALSE)
}
\arguments{
\item{n}{A vector of character factor levels (of length >=1) or the number
of unique levels (>= 1).}

\item{contrasts}{This argument is for backwards compatibility and only the
default of \code{TRUE} is supported.}

\item{sparse}{This argument is for backwards compatibility and only the
default of \code{FALSE} is supported.}
}
\value{
A diagonal matrix that is \code{n}-by-\code{n}.
}
\description{
This contrast function produces a model matrix with indicator columns for
each level of each factor.
}
\details{
By default, \code{model.matrix()} generates binary indicator variables for
factor predictors. When the formula does not remove an intercept, an
incomplete set of indicators are created; no indicator is made for the
first level of the factor.

For example, \code{species} and \code{island} both have three levels but
\code{model.matrix()} creates two indicator variables for each:

\if{html}{\out{<div class="sourceCode r">}}\preformatted{library(dplyr)
library(modeldata)
data(penguins)

levels(penguins$species)
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode">}}\preformatted{## [1] "Adelie"    "Chinstrap" "Gentoo"
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode r">}}\preformatted{levels(penguins$island)
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode">}}\preformatted{## [1] "Biscoe"    "Dream"     "Torgersen"
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode r">}}\preformatted{model.matrix(~ species + island, data = penguins) \%>\% 
  colnames()
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode">}}\preformatted{## [1] "(Intercept)"      "speciesChinstrap" "speciesGentoo"    "islandDream"     
## [5] "islandTorgersen"
}\if{html}{\out{</div>}}

For a formula with no intercept, the first factor is expanded to
indicators for \emph{all} factor levels but all other factors are expanded to
all but one (as above):

\if{html}{\out{<div class="sourceCode r">}}\preformatted{model.matrix(~ 0 + species + island, data = penguins) \%>\% 
  colnames()
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode">}}\preformatted{## [1] "speciesAdelie"    "speciesChinstrap" "speciesGentoo"    "islandDream"     
## [5] "islandTorgersen"
}\if{html}{\out{</div>}}

For inference, this hybrid encoding can be problematic.

To generate all indicators, use this contrast:

\if{html}{\out{<div class="sourceCode r">}}\preformatted{# Switch out the contrast method
old_contr <- options("contrasts")$contrasts
new_contr <- old_contr
new_contr["unordered"] <- "contr_one_hot"
options(contrasts = new_contr)

model.matrix(~ species + island, data = penguins) \%>\% 
  colnames()
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode">}}\preformatted{## [1] "(Intercept)"      "speciesAdelie"    "speciesChinstrap" "speciesGentoo"   
## [5] "islandBiscoe"     "islandDream"      "islandTorgersen"
}\if{html}{\out{</div>}}

\if{html}{\out{<div class="sourceCode r">}}\preformatted{options(contrasts = old_contr)
}\if{html}{\out{</div>}}

Removing the intercept here does not affect the factor encodings.
}