segmentation.OU.specialCase.lasso: Segmentation in the OU special case, using lasso regression

In PhylogeneticEM: Automatic Shift Detection using a Phylogenetic EM

Segmentation in the OU special case, using lasso regression

Description

segmentation.OU.specialCase.lasso performs the segmentation using a lasso regression to select for the edges where the shifts are added.

Usage

segmentation.OU.specialCase.lasso(
  phylo,
  nbr_of_shifts,
  D,
  Xp,
  penscale = rep(1, (nrow(phylo$edge) + 1)),
  ...
)

Arguments

phylo	a phylogenetic tree
nbr_of_shifts	Number of shifts on the phylogeny allowed
conditional_law_X	moments of the conditional law of X given Y, result of function compute_M.OU.specialCase
selection.strength	the selection strength

Details

This function re-write the sum of costs to be minimized as a least squares regression problem, and uses a lasso regression to solve it. It uses functions incidence.matrix.full to express the problem as a linear model.

Value

List containing : beta_0 : the optimal value at the root shifts : list containing the computed tau and delta costs : vector of costs

PhylogeneticEM documentation built on Aug. 31, 2022, 9:16 a.m.