specdens: Bootstrap for testing equality of spectral densities

In bootspecdens: Testing equality of spectral densities

Description

Bootsrap for testing the hypothesis that the spectral densities of a number m, m>= 2, not necessarily independent time series are equal. The test is based on a appropriate L2-distance measure between the nonparametric estimated individual spectral density and an overall pooled spectral density, obtained using the whole set of time series. The returned value is "equal spectral densities" for equal spectral densities respectively "not equal densities" for different spectral densities.

Usage

specdens(data, h, B, level)

Arguments

data	Data is a mxn matrix. In each row the data of one time series considered, i.e. m=number of time series and n=number of observation.
h	A single number which represents the kernel bandwidth smoothing parameter.
B	The number of bootstrap replicates. This will be a single positive integer. Normally something like 1000 is used.
level	A single number in (0,1) represents the level of the test. Normally 0.05 is used.

Value

The value returned is "equal spectral densities" for equal spectral densities respectively "not equal spectral densities" for different spectral densities. Furthermore the p-value is given.

Author(s)

Tatjana Kinsvater [e-mail]

References

Dette, H. and Paparoditis ,E. (2007) Testing Equality Of Spectral Densities.

Examples

data(Nile)
data(WWWusage)
data<- matrix(c(Nile[1:20],WWWusage[1:20]), nrow=2, byrow=TRUE) #creates the data matrix with 
										    #the first 20 values of the 
										    #time series Nile and WWWusage
specdens(data, 0.2, 100, 0.05)

bootspecdens documentation built on May 2, 2019, 3:22 a.m.