convergenceDiagnostics: Diagnose convergence of EMLasso.lognet objects

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

Description

Diagnose convergence of EMLasso.lognet objects

Usage

  convergenceDiagnostics(model, minIt = 10,
    checkConvergence = checkConvergence.glmnet, ...,
    verbosity = 0)

Arguments

model	model fit (should be the result of EMLasso)
minIt	check convergence from this iteration on
checkConvergence	method, normally checkConvergence.glmnet, that checks convergence.
...	passed on to checkConvergence
verbosity	The higher this value, the more levels of progress and debug information is displayed (note: in R for Windows, turn off buffered output)

Value

list with one item per lambda. Each item is a list with items:

numit	number of iterations until convergence for this lambda
simpleconv	for all iterations from minIt to the last one used, whether or not convergence occurred
changedNonZeroness	for all iterations from minIt to the last one used, the variable names for which zeroness is different from the last iteration

Author(s)

Nick Sabbe [e-mail]

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,
	lambdas=c(0.03,0.002,0.0003), nrOfSamplesPerMDRow=7, verbosity=2,
	convergenceChecker=convergenceCheckCreator(minIt=5, maxIt=10))
sfStop()
convergenceDiagnostics(iris.emlognet, minIt=2, verbosity=1)

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