TED (version 1.1.1)

eventDetection: Detect events from time series

Description

This function finds events from a time series.

Usage

eventDetection(x, w, noiseType = c("white", "red"), parallel = FALSE, alpha = 0.05, data = c("art", "real"))

Arguments

x

a vector or time series.

w

size of the sliding window.

noiseType

background noise type assumed for x. There are two options: white noise or red noise.

parallel

logical, if TRUE then codes are executed in parallel using foreach package. The user must register a parallel backend to use by the doMC package.

alpha

the significance level. When the noise test p value of the subsequence is smaller than this significance level, it is defined as a potential event. Default is 0.05.

data

type of data being analysed. There are two options: `art' if analysed data is artificial data and `real' if analysed data is real world turbulence data. Please see the details in Kang et al. (2014).

Value

an object of class 'events' with the components listed below:

x

the original time series.

start

a vector consisting of starting points of events.

end

a vector consisting of ending points of events.

nevents

number of detected events.

References

Yanfei Kang, Danijel Belusic, Kate Smith-Miles (2014): Detecting and Classifying Events in Noisy Time Series. J. Atmos. Sci., 71, 1090-1104. http://dx.doi.org/10.1175/JAS-D-13-0182.1.

Gregory S. Poulos, William Blumen, David C. Fritts, Julie K. Lundquist, Jielun Sun, Sean P. Burns, Carmen Nappo, Robert Banta, Rob Newsom, Joan Cuxart, Enric Terradellas, Ben Balsley, and Michael Jensen. CASES-99: A comprehensive investigation of the stable nocturnal boundary layer (2002). Bulletin of the American Meteorological Society, 83(4):555-581.

See Also

noiseTests, eventExtraction, plotevents

Examples

##################################
#   1st art eg (white noise)
##################################
set.seed(123)
n=128
types=c('box','rc','cr','sine')
shapes=matrix(NA,20,n)
for (i in 1:20){
  shapes[i,]=cbfs(type=types[sample(1:4,1)])
}
whitenoise=ts2mat(rnorm(128*20),128)
# generate x which randomly combine the four types of events with each two of them
# separated by noise
x=c(rnorm(128),t(cbind(shapes,whitenoise)))
plot(x,ty='l')
# specify a sliding window size and significant level
## Not run: 
# w=128; alpha=0.05
# events=eventDetection(x,w,'white',parallel=FALSE,alpha,'art')
# ## End(Not run)
##################################
#   2nd art eg (red noise)
##################################
set.seed(123)
# set a red noise level
coeff=0.5;s=1
# generated x with red noise as the background; this time series is the one used in
# Kang et al. (2014)
x=c(arima.sim(list(order = c(1,0,0),ar=coeff),n=500,sd=s),
    cbfs_red('rc'),arima.sim(list(order = c(1,0,0),ar=coeff),n=400,sd=s),
    cbfs_red('cr'),arima.sim(list(order = c(1,0,0),ar=coeff),n=400,sd=s),
    cbfs_red('box'),arima.sim(list(order = c(1,0,0),ar=coeff),n=400,sd=s),
    cbfs_red('sine'),arima.sim(list(order = c(1,0,0),ar=coeff),n=1000,sd=s),
    arima.sim(list(order = c(1,0,0),ar=0.8),n=1100,sd=4))
# specify a sliding window size and significant level
## Not run: 
# w=128; alpha=0.05
# # event detection
# events=eventDetection(x,w,'red',parallel=FALSE,alpha,'art')
# ## End(Not run)
##################################
#   CASES-99 dataset (9.5m)
##################################
# window size which needs to be chosen by the user
w=120
# specify a significant level
alpha=0.05
# event detection from CASES99 data
## Not run: 
# data(CASES99)
# CASESevents=eventDetection(CASES99,w,'red',parallel=FALSE,alpha,'real')
# ## End(Not run)