Skip to contents

To work with functional features, those features need to be stored as a matrix column in the data.frame, so mlr can automatically recognize them as functional features. This function allows for an easy conversion from a data.frame with numeric columns to the required format. If the data already contains matrix columns, they are left as-is if not specified otherwise in fd.features. See Examples for the structure of the generated output.

Usage

makeFunctionalData(data, fd.features = NULL, exclude.cols = NULL)

Arguments

data

(data.frame)
A data.frame that contains the functional features as numeric columns.

fd.features

(list)
Named list containing integer column indices or character column names. Each element defines a functional feature, in the given order of the indices or column names. The name of the list element defines the name of the functional feature. All selected columns have to correspond to numeric data.frame entries. The default is NULL, which means all numeric features are considered to be a single functional “fd1”.

exclude.cols

(character | integer)
Column names or indices to exclude from conversion to functionals, even if they are in included in fd.features. Default is not to exclude anything.

Value

(data.frame).

Examples

# data.frame where columns 1:6 and 8:10 belong to a functional feature
d1 = data.frame(matrix(rnorm(100), nrow = 10), "target" = seq_len(10))
# Transform to functional data
d2 = makeFunctionalData(d1, fd.features = list("fd1" = 1:6, "fd2" = 8:10))
# Create a regression task
makeRegrTask(data = d2, target = "target")
#> Supervised task: d2
#> Type: regr
#> Target: target
#> Observations: 10
#> Features:
#>    numerics     factors     ordered functionals 
#>           1           0           0           2 
#> Missings: FALSE
#> Has weights: FALSE
#> Has blocking: FALSE
#> Has coordinates: FALSE