compareTS: Community Time Series Comparison

Description Usage Arguments Details Value

View source: R/compareTS.R

Description

Compute properties of community time series generated with generateTS.

Usage

1
2
3
4
5
6
compareTS(input.folder = "", expIds = c(), modif.folder = "",
  modif = "", testNeutral = FALSE, sliceDef = c(1, NA), epsilon = 0.2,
  predef = FALSE, detrend = TRUE, norm = FALSE, hurstBins = c(0.5, 0.7,
  0.9), maxautocorBins = c(0.3, 0.5, 0.8), timeDecaySliceDef = c(1, 50),
  varEvolSliceDef = c(), radSliceDef = c(), returnDistribs = FALSE,
  returnTS = FALSE)

Arguments

input.folder

the folder where results (settings and time series) of function generateTS are stored.

expIds

the experiment identifiers of time series to be considered

modif.folder

the folder with time series sub-sets as generated by sliceTS.R or with noisy time series as generated by addNoise.R (settings are read in from input.folder)

modif

modification (shared name of time series in modif.folder, sliced for time series sub-sets and pois/multi for time series with Poisson or multinomial noise, respectively)

testNeutral

test neutrality (requires libraries WrightFisher and logitnorm)

sliceDef

consider the time series sub-set defined by the given start and end time point for property computation, an end time point of NA means that the entire time series is considered, a single value T in the sliceDef vector means that the last T time points are considered; timeDecaySliceDef and varEvolSliceDef refer to the outcome of sliceDef

epsilon

allowed deviation from the expected slope for noise type identification

predef

recognize noise types with predefined slope boundaries (see identifyNoisetypes for details)

detrend

remove linear trends before computing the periodogram (recommended)

norm

normalize time series by dividing each entry in a sample by the sum of its sample; if norm is true, entropy computation is omitted

hurstBins

binning thresholds for Hurst exponent (three thresholds required for four bins)

maxautocorBins

binning thresholds for maximal autocorrelation (three thresholds required for four bins)

timeDecaySliceDef

the time series subset considered to compute the time decay, a vector with the start and the end time point (if the end time point is NA, the entire time series is used)

varEvolSliceDef

(optional, by default not computed) the time series subset considered to compute the evolution of variance, a vector with the start and the end time point (if the end time point is NA, the entire time series is used)

radSliceDef

(optional, by default not computed) the time series subset considered to analyse the (non-normalized also when norm is enabled) species abundance versus rank curve, a vector with the start and the end time point (if the end time point is NA, the entire time series is used); cannot be used together with sliceDef

returnDistribs

return distributions at the final time point (no time series properties are computed, cannot be used together with returnTS)

returnTS

return the time series (no time series properties are computed, cannot be used together with returnDistribs)

Details

If infotheo is installed, the entropy will be also computed. Noise types are computed with smooth set to true. The neutrality test is carried out with ntests set to 500 and method logitnorm.

Value

a table with experiment parameters (algorithm, connectance, sigma, theta and so on) and time series properties (noise types percentages, slope of Taylor's law etc.); if returnDistrib is true, a list with the abundances at the last time point, if returnTS is true, a list with the time series


hallucigenia-sparsa/seqtime documentation built on June 16, 2018, 1:49 p.m.