tsmeasures: Computes a feature matrix of a set of time-series

In pli2016/anomalous: Unusual Time Series Detection

Description

Computes various different features (e.g., entropy, trend, seasonality) of a set of time-series

Usage

tsmeasures(y, normalise = TRUE, width = ifelse(frequency(y) > 1, frequency(y), 
           10), window = width) 

Arguments

y	A multivariate time series
normalise	If TRUE, each time series is scaled to be normally distributed with mean 0 and sd 1
width	A window size for variance change, level shift and lumpiness
window	A window size for KLscore

Value

An object of class features with the following components:

lumpiness	Variance of annual variances of remainder
entropy	Spectral entropy from ForeCA package
ACF1	First order of autocorrelation
lshift	Level shift using rolling window
vchange	Variance change
cpoints	The number of crossing points
fspots	Flat spots using disretization
trend	Strength of trend
linearity	Strength of linearity
curvature	Strength of curvature
spikiness	Strength of spikiness
season	Strength of seasonality
peak	Strength of peaks
trough	Strength of trough
KLscore	Kullback-Leibler score
change.idx	Index of the maximum KL score

Author(s)

Rob J Hyndman, Earo Wang, Nikolay Laptev

See Also

continuous_entropy

Examples

  y <- tsmeasures(dat0, window = 48)
  # Add mean and variance to the "features" matrix
  avg <- colMeans(dat0)
  variance <- apply(dat0, 2, var)
  y <- cbind(y, avg, variance)
  class(y) <- c("features", "matrix")

pli2016/anomalous documentation built on May 25, 2019, 8:22 a.m.