fitHurdle: Fit the hurdle model coordinate-by-coordinate on a sample

In amcdavid/HurdleNormal: 'Estimation and sampling for multi-dimensional Hurdle models on a Normal density with applications to single-cell co-expression'

Fit the hurdle model coordinate-by-coordinate on a sample

Description

Fit the hurdle model coordinate-by-coordinate on a sample

Usage

autoGLM(
  samp,
  fixed = NULL,
  parallel = FALSE,
  keepNodePaths = FALSE,
  checkpointDir = "ignored",
  nlambda = 200,
  lambda.min.ratio = 0.1,
  family = "binomial"
)

fitHurdle(
  samp,
  fixed = NULL,
  parallel = TRUE,
  keepNodePaths = FALSE,
  checkpointDir = NULL,
  makeModelArgs = NULL,
  indices,
  ...
)

Arguments

samp	matrix of data, columns are variables
fixed	data.frame of fixed covariates (to be conditioned upon)
parallel	parallelize over variables using "mclapply"?
keepNodePaths	return node-wise output (solution paths and diagnostics for each node) as attribute 'nodePaths'
checkpointDir	(optional) directory to save the fit of each node, useful for large problems. If it exists, then completed nodes will be automatically loaded.
nlambda	number of lambda values on grid (default 200).
lambda.min.ratio	minimum lambda ratio (as a function of lambda0, where the first predictor enters; default .1)
family	in the case of autoGLM one of "gaussian" or "logistic"
makeModelArgs	(optional) arguments passed to the model matrix function
indices	(optional) subset of indices to fit, useful for cluster parallelization.
...	passed to cgpaths

Value

list of fits, one per coordinate and an attribute "timing"

Functions

• autoGLM: Fit an auto-model (Ising or Gaussian) to samp using glmnet. checkpointDir is currently ignored.

See Also

neighborhoodToArray, autoGLM, interpolateEdges

amcdavid/HurdleNormal documentation built on May 14, 2022, 11:12 p.m.