nmcdr (version 0.3.0)

nmcd: Non-parametric Multiple Change-points Detection

Description

Detect multiple change-points using a non-parametric maximum likelihood approach.

Usage

nmcd(x, kmax, cpp, ncp = length(cpp), n)

Arguments

x

data vector

kmax

upper bound of the number of change-points

cpp

positions of candidate change-points. usually returned by functionisp. if this parameter is not assigned, the isp function will be called to assign it.

ncp

the number of the candidate change-points.

n

length of the data.

Value

a list with class nmcd is returned with elements:

npp

the true number of change-points

cpp

positions of true change-points

data

raw data, this is not printed on screen by default

bic

minimal BIC value gained.

Details

NMCD use DP algorithm to select change-points, while the true number of change-points is determined by the Bayesian information criterion(BIC).

References

Changliang Zou, Guosheng Yin, Long Feng, Zhaojun Wang. Non-parametric Maximum Likelihood Approach to Multiple Change-points Problem

Examples

t = nmcd(Blocks)
summary(t)
plot(t)