randomShuffleTest: Perform the random shuffle test

Description Usage Arguments Value Examples

View source: R/randomShuffleTest.r

Description

Perform the random shuffle test

Usage

1
2
randomShuffleTest(ts, its = NULL, blockLength = 1, p = 5, n = 1000,
  alpha = 0.05, returnExtraInfo = FALSE)

Arguments

ts

A time-series (a numeric vector)

its

Indices of which bin the values of x correspond to (a numeric vector of integers between 1 and p). This is optional, and taken to be cyclical by default.

blockLength

Shuffle the order of blocks of this length (a positive integer between 1 and less than length(ts) ). If blockLength is equal to 1, then the series is shuffled randomly. If blockLength is greater than 1, blocks of this length will be shuffled. The blockLength should be much less than length(ts) and should be co-prime with p.

p

Length of the period to test for (a positive integer)

n

Number of resamples to use in the random shuffle test (a positive integer)

alpha

P-value threshold to use when calculating the power (a positive scalar between 0 and 1, although it is recommended to be around 0.05)

returnExtraInfo

Logical value - if TRUE, the function will return extra information (the value of the range test statistic, the mean per bin, the departures within each bin from the mean value, and a confidence interval on the p-value). If FALSE, only the p-value is returned.

Value

This depends on the value of returnExtraInfo: - if returnExtraInfo is FALSE (the default), the function returns the p-value only - if returnExtraInfo is TRUE, the function returns a list with the p-value, the value of the range test statistic, the mean per bin, the departures within each bin from the mean value, and a confidence interval on the p-value

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
## Generate a random time-series:
##  - length = 1000
##  - signal period = 8
##  - signal to noise ratio = 0.2
set.seed(42)

x <- rep(rnorm(8)*0.2,length.out = 1000) + rnorm(1000)
randomShuffleTest(ts = x, p = 8, n = 1000,
                   blockLength = 10, returnExtraInfo = TRUE)

## compare with periods either too short or too long:
randomShuffleTest(ts = x, p = 7, n = 1000, blockLength = 10)
randomShuffleTest(ts = x, p = 9, n = 1000, blockLength = 10)

JeremySilver/RandomThinning documentation built on May 27, 2017, 12:02 a.m.