Parallel_PELTSM1: Parallel PELT SM1

In KayleaHaynes/changepoint.parallel: Parallel Changepoint Detection

Description

Splits the data into chunks and runs parallel changepoint detection on each chunk then merges the results and runs parallel detection using the detected changepoints in step 1 as the candidate changepoint locations.

Usage

Parallel_PELTSM1(data, penalty, pruning, sum.stat = norm.sum,
  cost = norm.mean.cost, ncores = 2, boundary = "fixed",
  boundary_value = 0, minseglen)

Arguments

data	A vector of data-points within which you with to find changepoints.
penalty	The value of the penalty.
pruning	If true PELT is used, if false Optimal Partitioning is used.
sum.stat	This can be “norm.sum" or “exp.sum".
cost	This can be “norm.mean.cost", “norm.var.cost", “norm.meanvar.cost" or “exp.cost".
ncores	Number of cores to use.
boundary	This can either be “fixed" or “adaptive".
boundary_value	If boundary is fixed then this is the number of points to use around the boundary.
minseglen	Minimim length a segment can be

Value

The detected changepoints.

Author(s)

Kaylea Haynes

Examples

### Set up parallel environment ###
library(doParallel)
library(foreach)

ncores <- c(10)
cl <- makeCluster(ncores)
registerDoParallel(cl)
### Generate some data from the blocks data set ####
cpts <- round(c(0,0.1, 0.13, 0.15, 0.23, 0.25, 0.40, 0.44, 0.65, 0.76, 0.78, 0.81,1)*10000)
segment_param <- c(0, 4, -1, 2, -2, 3, -1.2, 0.9, 5.2, 2.1, 4.2, 0)
data <- NULL

for (i in 1:(length(cpts)-1)){
  data_new <- rep(segment_param[i],cpts[i+1] - cpts[i])
 data <- c(data, data_new)
}
data <- data + rnorm(10000,0,1)

Parallel_PELTSM1(data, 2*log(length(data)), TRUE, sum.stat = norm.sum, cost = norm.mean.cost, ncores=10, boundary = "fixed",boundary_value = 20, 1)

KayleaHaynes/changepoint.parallel documentation built on May 7, 2019, 12:29 p.m.