hybrid_simultanee {segclust2d}

hybrid_simultanee performs a simultaneous seg - clustering for bivariate signals.

Description

It is an algorithm which combines dynamic programming and the EM algorithm to calculate the MLE of phi and T, which are the mixture parameters and the change point instants. this algorithm is run for a given number of clusters, and estimates the parameters for a segmentation/clustering model with P clusters and 1:Kmax segments

Usage

hybrid_simultanee(
  x,
  P,
  Kmax,
  lmin = 3,
  sameSigma = TRUE,
  sameVar.init = FALSE,
  eps = 1e-06,
  lissage = TRUE,
  pureR = FALSE,
  ...
)

Arguments

x

the two-dimensional signal, one line per dimension

P

the number of classes

Kmax

the maximal number of segments

lmin

minimum length of segment

sameSigma

should segment have the same variance

sameVar.init

sameVar.init

eps

eps

lissage

should likelihood be smoothed

pureR

should algorithm run in full R or use Rcpp speed improvements

...

additional parameters

Value

a list with Linc, the incomplete loglikelihood =Linc,param=paramtau posterior probability