mlr: extractFDATsfeatures – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

extractFDATsfeatures

Time-Series Feature Heuristics

Description

The function extracts features from functional data based on known Heuristics. For more details refer to tsfeatures::tsfeatures(). Under the hood this function uses the package tsfeatures::tsfeatures(). For more information see Hyndman, Wang and Laptev, Large-Scale Unusual Time Series Detection, ICDM 2015.

Note: Currently computes the following features:
"frequency", "stl_features", "entropy", "acf_features", "arch_stat", "crossing_points", "flat_spots", "hurst", "holt_parameters", "lumpiness", "max_kl_shift", "max_var_shift", "max_level_shift", "stability", "nonlinearity"

Usage

extractFDATsfeatures(
  scale = TRUE,
  trim = FALSE,
  trim_amount = 0.1,
  parallel = FALSE,
  na.action = na.pass,
  feats = NULL,
  ...
)

Arguments

`scale`	(`logical(1)`) If TRUE, time series are scaled to mean 0 and sd 1 before features are computed.
`trim`	(`logical(1)`) If TRUE, time series are trimmed by `trim_amount` before features are computed. Values larger than trim_amount in absolute value are set to NA.
`trim_amount`	(`numeric(1)`) Default level of trimming if `trim==TRUE`.
`parallel`	(`logical(1)`) If `TRUE`, multiple cores (or multiple sessions) will be used. This only speeds things up when there are a large number of time series.
`na.action`	(`logical(1)`) A function to handle missing values. Use `na.interp` to estimate missing values
`feats`	(`character`) A character vector of function names to apply to each time-series in order to extract features. Default: feats = c("frequency", "stl_features", "entropy", "acf_features", "arch_stat", "crossing_points", "flat_spots", "hurst", "holt_parameters", "lumpiness", "max_kl_shift", "max_var_shift", "max_level_shift", "stability", "nonlinearity")
`...`	(any) Further arguments passed on to the respective tsfeatures functions.

Value

(data.frame)

References

Hyndman, Wang and Laptev, Large-Scale Unusual Time Series Detection, ICDM 2015.

Bernd Bischl [aut] (<https://orcid.org/0000-0001-6002-6980>), Michel Lang [aut] (<https://orcid.org/0000-0001-9754-0393>), Lars Kotthoff [aut], Patrick Schratz [aut, cre] (<https://orcid.org/0000-0003-0748-6624>), Julia Schiffner [aut], Jakob Richter [aut], Zachary Jones [aut], Giuseppe Casalicchio [aut] (<https://orcid.org/0000-0001-5324-5966>), Mason Gallo [aut], Jakob Bossek [ctb] (<https://orcid.org/0000-0002-4121-4668>), Erich Studerus [ctb] (<https://orcid.org/0000-0003-4233-0182>), Leonard Judt [ctb], Tobias Kuehn [ctb], Pascal Kerschke [ctb] (<https://orcid.org/0000-0003-2862-1418>), Florian Fendt [ctb], Philipp Probst [ctb] (<https://orcid.org/0000-0001-8402-6790>), Xudong Sun [ctb] (<https://orcid.org/0000-0003-3269-2307>), Janek Thomas [ctb] (<https://orcid.org/0000-0003-4511-6245>), Bruno Vieira [ctb], Laura Beggel [ctb] (<https://orcid.org/0000-0002-8872-8535>), Quay Au [ctb] (<https://orcid.org/0000-0002-5252-8902>), Martin Binder [ctb], Florian Pfisterer [ctb], Stefan Coors [ctb], Steve Bronder [ctb], Alexander Engelhardt [ctb], Christoph Molnar [ctb], Annette Spooner [ctb]

Initial release