# onehotencode

Encode data labels into one-hot vectors

## Syntax

## Description

encodes data labels in categorical array `B`

= onehotencode(`A`

,`featureDim`

)`A`

into a one-hot encoded array
`B`

. The function replaces each element of `A`

with a
numeric vector of length equal to the number of unique classes in `A`

along
the dimension specified by `featureDim`

. The vector contains a
`1`

in the position corresponding to the class of the label in
`A`

, and a `0`

in every other position. Any
`<undefined>`

values are encoded to `NaN`

values.

encodes categorical data labels in table `tblB`

= onehotencode(`tblA`

)`tblA`

into a table of one-hot
encoded numeric values. The function replaces the single variable of `tblA`

with as many variables as the number of unique classes in `tblA`

. Each row in
`tblB`

contains a `1`

in the variable corresponding to
the class of the label in `tlbA`

, and a `0`

in all other
variables.

`___ = onehotencode(___,`

encodes the labels into numeric values of data type `typename`

)`typename`

. Use this syntax with any of the input and output arguments in previous syntaxes.

`___ = onehotencode(___,'ClassNames',`

also specifies the names of the classes to use for encoding. Use this syntax when
`classes`

)`A`

or `tblA`

does not contain categorical values,
when you want to exclude any class labels from being encoded, or when you want to encode the
vector elements in a specific order. Any label in `A`

or
`tblA`

of a class that does not exist in `classes`

is
encoded to a vector of `NaN`

values.

## Examples

## Input Arguments

## Output Arguments

## Alternative Functionality

To encode data labels, you can also use `dummyvar`

, which creates dummy variables from grouping variables. The following
table compares the `onehotencode`

and `dummyvar`

functions for different use cases.

Use Case | When to Use `onehotencode` | When to Use `dummyvar` |
---|---|---|

Encoding multiple variables | Use `onehotencode` in a loop. For an example, see One-Hot Encode Table with Several Variables. | Specify the input argument `group` as a cell array or positive
integer matrix. For examples, see Create Dummy Variables from Multiple Grouping Variables and Create Dummy Variables from Numeric Grouping Variables. |

Encoding a variable in cell array format | Convert the cell array variable to a categorical array. | Specify the input argument `group` as a cell array containing
one or more grouping variables. |

Encoding noncategorical data labels | Specify the data labels as a categorical array or specify the classes to encode
using the `ClassNames` name-value argument. For an example, see
One-Hot Encode Subset of Classes. | You do not need to convert the data labels, because `dummyvar`
accepts noncategorical grouping variables as input. |

Encoding an array of data labels | Specify the dimension to expand (`featureDim` ). | The software automatically determines the dimension to expand.
`dummyvar` returns dummy variables as a numeric array with
columns created from the columns of the input grouping variables. |

In many cases, you do not need to use the `onehotencode`

or
`dummyvar`

function for encoding. Most Statistics and Machine Learning Toolbox™ functions can operate directly on categorical response data. Most classification
and regression functions also accept categorical predictors.

## Version History

**Introduced in R2021b**