getMinMaxPosLikeGlmnet: Find a region of interest in a set of lambdas given criteria...
In EMLasso: Fit (logistic) LASSO when there is missing data in the predictors

Description Usage Arguments Value Author(s) Examples

Find a region of interest in a set of lambdas given criteria and their SE

  getMinMaxPosLikeGlmnet(object, ...)

  ## Default S3 method:
getMinMaxPosLikeGlmnet(object, cvm,
    cvsd, type.measure,...)

  ## S3 method for class 'cv.glmnet'
 getMinMaxPosLikeGlmnet(object, ...)

`object`	(main) object to get the region information from (for `default`: vector of lambda values)
`...`	for flexibility in 'derived' implementation
`cvm`	criteria value (see `cv.glmnet`)
`cvsd`	criteria value standard errors (see `cv.glmnet`)
`type.measure`	see `cv.glmnet`. If this is "auc", the `cvm` are negated before continuing.

list of class "lambdaregion":

`pos.optimum`	position of the optimum criterion in the set of lambdas
`pos.higherlambda`	position in the set of lambdas of the highest lambda with criterion within one se of the optimum
`pos.lowerlambda1`	position in the set of lambdas of the lambda below that of the optimum, that has the inverted optimum
`pos.lowerlambda2`	position in the set of lambdas of the highest lambda below that of the optimum that does not have its criterion within one se of the optimum
`lambda`	vector of lambda values (as passed in)

Nick Sabbe nick.sabbe@ugent.be

y<-sample(0:1, nrow(iris), replace=TRUE)
cvl<-fit.lognet(ds=iris, out=y, lambda=NULL, verbosity=10, type.measure="auc")
getMinMaxPosLikeGlmnet(cvl$lambda, cvl$cvm, cvl$cvsd, type.measure="auc")
y<-sample(0:1, nrow(iris), replace=TRUE)
cvl<-fit.lognet(ds=iris, out=y, lambda=NULL, verbosity=10, type.measure="auc")
getMinMaxPosLikeGlmnet(cvl)