missing.control {prioritylasso}

Construct control structures for handling of missing data for prioritylasso

Description

Construct control structures for handling of missing data for prioritylasso

Usage

missing.control(
  handle.missingdata = c("none", "ignore", "impute.offset"),
  offset.firstblock = c("zero", "intercept"),
  impute.offset.cases = c("complete.cases", "available.cases"),
  nfolds.imputation = 10,
  lambda.imputation = c("lambda.min", "lambda.1se"),
  perc.comp.cases.warning = 0.3,
  threshold.available.cases = 30,
  select.available.cases = c("maximise.blocks", "max")
)

Arguments

handle.missingdata

how blockwise missing data should be treated. Default is none which does nothing, ignore ignores the observations with missing data for the current block, impute.offset imputes the offset for the missing values.

offset.firstblock

determines if the offset of the first block for missing observations is zero or the intercept of the observed values for handle.missingdata = ignore

impute.offset.cases

which cases/observations should be used for the imputation model to impute missing offsets. Supported are complete cases (additional constraint is that every observation can only contain one missing block) and all available observations which have an overlap with the current block.

nfolds.imputation

nfolds for the glmnet of the imputation model

lambda.imputation

which lambda-value should be used for predicting the imputed offsets in cv.glmnet

perc.comp.cases.warning

percentage of complete cases when a warning is issued of too few cases for the imputation model

threshold.available.cases

if the number of available cases for impute.offset.cases = available.cases is below this threshold, prioritylasso tries to reduce the number of blocks taken into account for the imputation model to increase the number of observations used for the imputation model.

select.available.cases

determines how the blocks which are used for the imputation model are selected when impute.offset.cases = available.cases. max selects the blocks that maximise the number of observations, maximise.blocks tries to include as many blocks as possible, starting with the blocks with the hightes priority

Value

list with control parameters