EMLasso: Fit EMLasso to a complete dataset, creating a GLoMo for each...

In EMLasso: Fit (logistic) LASSO when there is missing data in the predictors

Description

Fit EMLasso to a complete dataset, creating a GLoMo for each lambda

Usage

  EMLasso(ds, out,
    family = c("gaussian", "binomial", "poisson", "multinomial", "cox"),
    lambdas = smartLambdaRetriever(),
    imputeDs2FitDsProperties = normalImputationConversion(),
    ..., verbosity = 0, logdir = "./",
    saveTempResults = TRUE,
    postProcessAll = postProcessEMLasso)

Arguments

ds	dataset with predictors
out	vector (binary factor) of outcomes
family	see glmnet
lambdas	see dfrConversionProps
imputeDs2FitDsProperties	see imputeDs2FitDs object that will provide the conversion from imputed dataset to one that is ready for fitting the predictor model
...	passed on to EMLasso.1l
verbosity	The higher this value, the more levels of progress and debug information is displayed (note: in R for Windows, turn off buffered output)
logdir	path to folder where logfiles (and results) of each repeat are stored
saveTempResults	if TRUE, save the results of each (parallelized)
postProcessAll	function, like postProcessEMLasso (its default) and of that signature, to do some extra work before returning the result of EMLasso

Value

an object of class "EMLasso", "glmnet" and some other classes based on the class of the outcome model. The items added to the glmnet members are:

result	list of EMLasso1l objects per lambda
params	EMLasso.1l.param parameters passed in
logdir	directory where logging/saving occurred
combinedGLoMo	GLoMo object: the combination (through combineGLoMos) of the GLoMos for each lambda (note: this item is added by postProcessEMLasso, so it may not be present if another postProcessAll is used)
imputeDs2FitDsProperties	as passed in, but first fed to imputeDs2FitDsProps
family	as passed in

Note

If lambdas is not passed along or is NULL, a set of lambdas is used by utilizing findReasonableLambdaHelper

Author(s)

Nick Sabbe [e-mail]

See Also

EMLasso.1l

Examples

y<-rbinom(nrow(iris), 1, 0.5)
require(addendum)
require(NumDfr)
require(GLoMo)
require(snowfall)
require(EMLasso)
sfInit(parallel = FALSE, cpus = 1)
sfLibrary(addendum)
sfLibrary(NumDfr)
sfLibrary(GLoMo)
sfLibrary(EMLasso)
iris.cpy<-randomNA(iris, n=0.1)
iris.emlognet<-EMLasso(ds=numdfr(iris.cpy), out=y, family="binomial",
	lambdas=c(0.03,0.002,0.0003), nrOfSamplesPerMDRow=7, verbosity=2,
	convergenceChecker=convergenceCheckCreator(minIt=5, maxIt=10))
sfStop()

EMLasso documentation built on May 2, 2019, 5:49 p.m.