collectImputationModels: Collect data from a model, needed for multiple imputation...

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

Description

Collect data from a model, needed for multiple imputation based on this model

Usage

  collectImputationModels(model, ds = model$result[[1]]$ds,
    ..., verbosity = 0)

  ## S3 method for class 'EMLassoGLoMo'
collectImputationModels(model,
    ds=model$result[[1]]$ds, useCombinedGLoMo=TRUE, ...,
    verbosity=0)

  ## S3 method for class 'EMLassoGLoMoImputationData'
predict(object,
    newdata, out, wts=rep(1, nrow(newdata)),
    type.measure="auc", actualPredictAndEvaluateFunction,
    unpenalized=FALSE, ..., verbosity=0)

Arguments

model	model fit
ds	dataset for which imputation will need to happen
...	implementation dependent. e.g. for EMLassoGLoMo should be a list containig an item useCombinedGLoMo (will default to TRUE); for predict.EMLassoGLoMoImputationData: passed on to cv.glmnet. Note: the ones implied by object are not supported! e.g. family and standardize cannot be overridden
verbosity	The higher this value, the more levels of progress and debug information is displayed (note: in R for Windows, turn off buffered output)
useCombinedGLoMo	if FALSE (not the default), a distinct GLoMo is used for every lambda. Otherwise, the combined GLoMo (see combineGLoMos) is used for all lambdas.
object	"EMLassoGLoMoImputationData" that holds the information to perform imputations
newdata	dataset for which imputation needs to occur
out	outcomes that will be used for evaluating the models
wts	weights per observation (defaults to equal weights for all observations)
type.measure	see cv.glmnet
actualPredictAndEvaluateFunction	function similar to the nonexported .predAndEvalGLN function. Expected to support parameters: useGLoMo, useReusable, useLambda, newdata, out, weights, type.measure, imputeDs2FitDsProperties, nobs, ..., verbosity=0. Should return a vector twice the length of the number of rows: first all evaluated values, then all their sds. Note: if this is not specified, .predAndEvalGLN is used.
unpenalized	if TRUE (not the default) a simple regression model is fit with the selected variables

Value

collectImputationModels.EMLassoGLoMo will return an object of class "EMLassoGLoMoImputationData". This is a list with items:

imputationParams	the ... passed along
useCombinedGLoMo	as passed along
combinedGLoMo	if relevant (useCombinedGLoMo==TRUE): the combined GLoMo
glomolist	if relevant (useCombinedGLoMo==FALSE): list containing the GLoMo to be used for each lambda
reusableData	either a single object of class "ReusableDataForGLoMoSampling" (see reusableDataForGLoMoSampling), or a list with one of them per lambda
lambda	taken from model
family	taken from model
imputeDs2FitDsProperties	taken from model

Note

aids to generalize crossvalidation

Author(s)

Nick Sabbe [e-mail]

See Also

EMLasso, cv.glmnet

EMLasso, cv.glmnet

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