simplyValidate: Validate a model (to its original data)
In EMLasso: Fit (logistic) LASSO when there is missing data in the predictors

Description Usage Arguments Value Note Author(s) See Also Examples

Validate a model (to its original data)

  simplyValidate(model, ..., verbosity = 0)

  ## S3 method for class 'EMLassoGLoMo'
simplyValidate(model,
    ds=model$result[[1]]$ds, out=model$result[[1]]$out,
    wts=rep(1, nrow(ds)),
    imputeDs2FitDsProperties=model$imputeDs2FitDsProperties,
    imputations=10, ..., type.measure="auc",
    keepResultPerLambda=FALSE, nobs=1, unpenalized=FALSE,
    verbosity=0)

`model`	model fit
`...`	for flexibility in 'derived' implementation
`verbosity`	The higher this value, the more levels of progress and debug information is displayed (note: in R for Windows, turn off buffered output)
`ds`	dataset with predictors
`out`	vector (binary factor) of outcomes
`wts`	vector of weights (defaults to equal weights for all rows)
`imputeDs2FitDsProperties`	see `imputeDs2FitDs` object that will provide the conversion from imputed dataset to one that is ready for fitting the predictor model
`imputations`	Number of multiple imputations on the complete dataset (defaults to 10)
`type.measure`	see `cv.glmnet`
`keepResultPerLambda`	if `TRUE` (not the default), the individual results from the `simplyValidate.EMLasso1l` are also returned in an extra item `resultPerLambda`
`nobs`	how many observations are simulated for each row with missing data
`unpenalized`	if `TRUE` (not the default) a simple regression model is fit with the selected variables

object that has as class: "sv." pasted before the class of model. Normally, model will will be the return value of EMLasso, so this result is mainly the same as a cv.glmnet. The added/altered items are:

`glmnet.fit`	is now the `model` passed in, so has more classes besides "glmnet" (e.g. "EMLasso")
`resultPerLambda`	matrix with one column per imputation. The top rows are the estimates for the criterion per lambda, below that are their SD estimates. Not present if `keepResultPerLambda=FALSE`

aids to generalize validation

Nick Sabbe nick.sabbe@ugent.be

EMLasso, cv.glmnet

y<-rbinom(nrow(iris), 1, 0.5)
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()
iris.sv.emlognet<-simplyValidate(iris.emlognet, verbosity=2)