func_dc: Double CUSUM statistics

In factorcpt: Simultaneous Change-Point and Factor Analysis

Description

Aggregates high-dimensional CUSUM statistic series via Double CUSUM approach.

Usage

func_dc(z)

Arguments

z	input data matrix

Value

cs	a matrix of CUSUM series
acs	a matrix of absolute values of CUSUM series
res	pointwise maximum of Double CUSUM statistics
mat	a matrix of Double CUSUM statistics

Author(s)

Haeran Cho

References

H. Cho (2016) Change-point detection in panel data via double CUSUM statistic. Electronic Journal of Statistics. 10: 2000-2038.

Examples

e <- matrix(rnorm(100*100), nrow=100)
f <- matrix(0, nrow=100, ncol=100)
f[, 51:100] <- 1 # t = 50 represents the change-point in the means
x <- f + e
fd <- func_dc(x)
plot(fd$res, type='l'); abline(v=50, col=2) # pointwise maximum of DC statistics
image(t(fd$mat)) # heatmap of DC statistics with x-axis representing the time

Example output

factorcpt documentation built on May 2, 2019, 8:15 a.m.