trend {utsf}R Documentation

Specifying the transformation for dealing with trended series

Description

This function is used to specify the preprocessing associated with the trend of a time series.

Usage

trend(
  type = c("additive", "multiplicative", "differences", "none"),
  n = -1,
  transform_features = TRUE
)

Arguments

type

A character indicating the type of preprocessing applied to the time series. Possible values are: "none", "additive", "multiplicative" and "differences".

n

An integer specifying the order of first differences to be applied. If the default (-1) is used, the order of first differences needed by the time series will be estimated by the forecast::ndiffs() function. This parameter is only meant when the type parameter is "differences".

transform_features

A logical value indicating whether the training features are also transformed with the additive or multiplicative transformation.

Value

A list with the selected options

Examples

trend("none")            # no preprocessing
trend("additive")        # additive preprocessing
trend("differences", 1)  # order 1 first differences
trend("differences", -1) # order of first differences automatically estimated

[Package utsf version 1.3.0 Index]