Extract the imputed datasets from a multiple-imputation object
Source:R/result_accessors.R
get_imputed_datasets.RdReturns the list of completed datasets stored in a multiple-imputation
object, equivalent to mi$datasets.
Arguments
- mi
A
pigauto_miobject frommulti_impute(),multi_impute_analysis(), ormulti_impute_trees().
Value
A list of mi$m input-shaped data frames. Observed cells are
preserved; missing trait cells vary across datasets according to the
draw mechanism. Check mi$draws_method before analysing them:
"posterior" draws support the documented pooling workflow via
with_imputations() and pool_mi(), while "conformal" and
"mc_dropout" draws are prediction diagnostics that describe the
spread of plausible imputed values and are refused by the pooling
functions. Datasets from multi_impute_trees() are tree-sensitivity
diagnostics; mi$tree_index records the tree behind each one.
See also
multi_impute(), with_imputations(), pool_mi(), and
completed_data() for the single completion returned by impute().
Examples
if (FALSE) { # \dontrun{
mi <- multi_impute(traits, tree, m = 20)
datasets <- get_imputed_datasets(mi)
} # }