SKNv0.2.0

Bootstrap tests for symmetry, kurtosis, and normality under dependence.

DOC 01 / 06Research software

Bootstrap tests for skewness, kurtosis, and normality

An R package for studentized double-bootstrap and non-studentized single-bootstrap inference under unspecified short or long memory.

Abstract

SKN provides autoregressive-sieve bootstrap tests for symmetry, mesokurtosis, and moment conditions associated with normality in a stationary univariate time series. The procedures accommodate antipersistence, short memory, and long memory without requiring the memory parameter to be estimated first.

Primary functionSKNtest()
Test familySample skewness, robust skewness, excess kurtosis, and combined normality
Default routeStudentized statistic with a double bootstrap
Alternative routeNon-studentized statistic with a single bootstrap
AuthorsCharisios Grivas, Zacharias Psaradakis, and Marián Vávra
ContributorsAlex Toft Gregersen and Lisa Rijkhoff
Read the methodsCompare bootstrap routesNile application
Minimal use

One interface, two bootstrap routes

The studentized route is the package default. Set studentized = FALSE to select the computationally lighter single-bootstrap route.

R / studentized default
library(SKN)
data("NileMin", package = "longmemo")

result <- SKNtest(
  NileMin,
  B1 = 1000,
  B2 = 200,
  crit = "AIC",
  studentized = TRUE
)
result
Computation: Production bootstrap counts can be expensive. This default double-bootstrap call is shown without being run during the website build.
Dependence

One procedure across memory regimes

The autoregressive sieve approximates the dependence structure with a finite-order autoregression. The same testing framework can therefore be used under antipersistence, short memory, or long memory without first selecting a memory regime.