kza (version 2.03)

kz_decompose: Kolmogorov-Zurbenko decompose

Description

Kolmogorov-Zurbenko decomposition of univariate time series into long term and seasonal components.

Usage

kz_decompose(y, k=3, period=NULL)
## S3 method for class 'kzd':
fitted(object, \dots)
## S3 method for class 'kzd':
residuals(object, \dots)
## S3 method for class 'kzd':
trend(object, \dots)
## S3 method for class 'kzd':
frequency(x, \dots)
## S3 method for class 'kzd':
predict(object, n.ahead = 1, prediction.interval = FALSE, level = 0.95, \dots)
## S3 method for class 'kzd':
plot(x, main = paste("KZ Decomposition of time series"), ...)
## S3 method for class 'kzd':
summary(object, digits=getOption("digits"), ...)

Arguments

y

univariate time series.

k

Number of iterations.

period

Vector with the window sizes for the long term and seasonal period.

object

kzd object returned from kz_decompose.

x

kzd object returned from kz_decompose.

...

Other parameters.

n.ahead

The number of data points ahead to predict.

prediction.interval

prediction interval level

level

level

digits

precision

main

Plot title.

Details

KZ is an iterated moving average filter.

Examples

yrs <- 20
t <- seq(0,yrs,length=yrs*365)
amp<-10
y <- amp*sin(2*pi*t*12+10)
y<-ts(y,start=c(1980,10),frequency=365)
signal<-y

tr<-seq(1, amp/2, length.out=7300)
y.orig<-signal+exp(tr)+300
y<-y.orig
set.seed(6); e <- rnorm(n = length(t), sd = 50)
SNR<-mean(abs(y)^2)/(mean((abs(e)^2)))
y<-log(y+e)

a<-kz_decompose(y, period=c(365,30))
plot(a)