Extracts the posterior distributions of missing values that were imputed by the model.
Arguments
- object
A
becausemodel object.- id_col
Optional character vector of IDs (e.g., species names) corresponding to the rows of the data. If the data in the model object already has names (e.g., from a phylogenetic model), these are used automatically. If provided, this vector must have the same length as the number of observations in the model (N).
Value
A data frame containing:
Variable: Name of the variable with missing data.ID: The identifier (e.g., Species) for the observation (if available).RowIndex: The original row index in the data.Mean: Posterior mean of the imputed value.SD: Posterior standard deviation.Q2.5: 2.5% quantile (lower credible interval).Q50: Median.Q97.5: 97.5% quantile (upper credible interval).
Details
This function identifies which values in the original data were missing (NA)
and looks up the corresponding imputed nodes in the posterior samples.
Note: The imputed values are only available if they were monitored during the run.
You must use monitor = "all" (or include the variable names in monitor)
when running because() to ensure these values are saved.
Examples
if (FALSE) { # \dontrun{
# Assuming 'fit' is a because model run with missing data and monitor="all"
imputed_values <- extract_imputed(fit)
head(imputed_values)
} # }