GA: Genetic algorithm

In changepointGA: Changepoint Detection via Modified Genetic Algorithm

Genetic algorithm

Description

Perform the modified genetic algorithm for changepoint detection. This involves the minimization of an objective function using a genetic algorithm (GA). The algorithm can be run sequentially or with explicit parallelization.

Usage

GA(
  ObjFunc,
  N,
  GA_param = .default.GA_param,
  GA_operators = .default.operators,
  p.range = NULL,
  ...
)

Arguments

ObjFunc	The fitness function to be minimized. Users can specify any R or Rcpp function as the fitness function, setting the input as the potential solution to the optimization problem and returning a numerical value as the output/fitness. Depending on the user-specified chromosome representation, the optimization task can be changepoint detection only or changepoint detection plus model order selection, which can be specified via the option parameter in GA_param. When option="both", the list p.range must be specified to give the range of model orders.
N	The sample size of the time series.
GA_param	A list contains the hyper-parameters for genetic algorithm. The default values for these hyper-parameters are included in .default.GA_param. See GA_param for the details.
GA_operators	A list includes the functions for population initialization, new individual selection, and genetic operator of crossover and mutation. See operators for the details and default functions.
p.range	Default is NULL for only changepoint detection. If p.range is specified as a list object, which contains the range of each model order parameters for order selection (integers). The number of order parameters must be equal to the length of p.range.
...	additional arguments that will be passed to the fitness function.

Value

Returns a list that has components:

overbestfit	The obtained minimum value of the objective function after the final iteration.
overbestchrom	The locations of the detected changepoints associating with the overbestfit the after the final iteration.
bestfit	The minimized fitness function values at each iteration.
bestchrom	The detected changepoints at each iteration.
count	The number of iterations undertaken by the genetic algorithm.
convg	An integer code. 0 indicates the algorithm successful completion. 1 indicates the the total number of generations exceeds the prespecified maxgen limit.

Examples

N = 1000
XMatT = matrix(1, nrow=N, ncol=1)
Xt = ts.sim(beta=0.5, XMat=XMatT, sigma=1, phi=0.5, theta=NULL,
            Delta=c(2, -2), CpLoc=c(250, 750), seed=1234)
TsPlotCheck(X=1:N, Xat=seq(from=1, to=N, length=10), Y=Xt, tau=c(250, 750))

GA.res = GA(ObjFunc=BinSearch.BIC, N=N, Xt=Xt)
GA.res$overbestfit
GA.res$overbestchrom

changepointGA documentation built on April 4, 2025, 4:39 a.m.