smash_gen_pois: Smooth Poisson sequence, accounting for nugget effect
In DongyueXie/smashrgen: Empirical Bayes smoothers

smash_gen_pois

R Documentation

Smooth Poisson sequence, accounting for nugget effect

Description

Smooth Poisson sequence, accounting for nugget effect

Usage

smash_gen_pois(
  x,
  s = 1,
  nug.init = NULL,
  est_nugget = TRUE,
  transformation = "lik_expan",
  lik_expan_at = "logx",
  nug.est.limit = 1,
  smoother = "smash",
  robust = FALSE,
  robust.q = 0.99,
  ash_pm_init_for0 = TRUE,
  eps = "estimate",
  filter.number = 1,
  family = "DaubExPhase",
  homoskedastic = FALSE,
  est_nugget_maxiter = 2,
  est_nugget_tol = 0.01
)

Arguments

`x`	observed Poisson sequence
`s`	Scale factor for Poisson observations: y~Pois(scale*lambda), can be a vector.
`nug.init`	init value of nugget effect, either a scalar or NULL
`transformation`	transformation of Poisson data, either 'vst' or 'lik_expan'; 'vst' for variance stabilizing transformation; 'lik_expansion' for likelihood expansion
`lik_expan_at`	if transformation='lik_expan', where to expand it? Can be logx, or smash_poi
`smoother`	smoothing method for Gaussian sequence, either 'smash' or 'ti.thresh'. When n is large, ti.thresh is much faster
`robust`	whether perform robust wavelet regression
`robust.q`	quantile to determine outliers
`eps`	If choose lik_expansion, if x=0, set x = x + eps. Either input a numerical value or 'estimate'. If estimate, eps = sum(x==1)/sum(x<=1)
`filter.number, family`	wavelet basis, see wavethresh package for more details
`ash.pm`	If choose lik_expansion, whether use ash posterior mean approximation if x=0. If not x = x+eps.
`maxiter`	max iterations for estimating nugget effect
`tol`	tolerance to stop iterations.