LOO_CV_with_PE: Leave-one-out cross-validation under pulsed evolution model

In wu-lab-uva/RasperGade: Reconstructing Ancestral State under Pulsed Evolution in R by Gaussian Decomposition

Leave-one-out cross-validation under pulsed evolution model

Description

Predict hidden states under the pulsed evolution model in a leave-one-out cross-validation

Usage

crossValidationWithPE(
  FMR,
  add.epsilon = TRUE,
  laplace = FALSE,
  numApprox = 1,
  margin = 1e-06,
  numCores = 1,
  asymptotic = 5
)

LOO_CV_with_PE(
  FMR,
  add.epsilon = TRUE,
  laplace = FALSE,
  numApprox = 1,
  margin = 1e-06,
  numCores = 1,
  asymptotic = 5
)

Arguments

FMR	the returned data structure from function fullMarginalReconstructionWithPE
add.epsilon	logical, if true, time-independent variation is added to the variance in the data frame
laplace	logical, if true, Laplace distribution is used for time-independent variation
numApprox	the number of normal distributions to approximate the Laplace distribution
margin	the total probability mass that the number of jumps omitted in a compound Poisson process
numCores	the number of cores to run in parallel
asymptotic	the threshold of expected number of jumps on a branch beyond which normal distribution is assumed

Value

$summary is a data frame listing the means and variances of the hidden states

$error is a list of error distributions where each element is a data frame

wu-lab-uva/RasperGade documentation built on June 24, 2022, 2:47 p.m.